You're unlikely to keep up with AI working in isolation. Your learning compounds faster when it's shared. That's the flywheel principle.

What’s a Flywheel?

A flywheel is a heavy rotating disk — it takes effort to get moving, but once it's spinning, momentum carries it. In organizational terms, a flywheel is any self-reinforcing loop where each action feeds the next. Share something you learned → a colleague tries it → they share what they found → you learn something you wouldn't have discovered on your own → you share that. The wheel spins faster with less effort over time.

The External Flywheel: Plug into People Who are Already Sharing

The fastest way to start is to follow a few people whose flywheels are already spinning and whose work is aligned with yours. You don't need to follow everyone. Pick a couple of people who speak your language on a platform you resonate with. Here are the people I follow and share about the most:

• Ethan Mollick is a Wharton professor. I was so thankful to see the rigor and focus he brings to his newsletter. The pressure was off me!

• Alexandra Samuel covers AI and the future of work for the Wall Street Journal and Harvard Business Review; her biweekly "Thrive at Work" newsletter is one of the cleanest signals available.

• If learning and development is your area, Philippa Hardman connects the science of learning with AI in practical, evidence-based ways

• Phil Simon writes about AI tools. At the moment, with a focus on Claude. He offers a hands-on site that connects technical topics to pop culture.

Pick two from my list or have one of the AIs help you find your fav. Subscribe. When something lands, don't just file it. Share it with a colleague.

The Internal Flywheel: Your Team Sharing with Each Other

The external flywheel feeds you. The internal one is where the real organizational learning happens. This is the harder build: it requires people to share things that aren't polished, to ask questions they're not sure are "smart," or to talk about strategies that failed. Take a look at Prof. Amy Edmondson’s work to see why.

Start Your Team’s Flywheel this Week… or give one you have a bit of a push. That power is exactly what my brother fears as he sits in that teacup.

The Pace Is Not Slowing

Look at this chart from the Stanford HAI 2025 AI Index. The dashed line is human-level performance. One by one, AI has surpassed human capabilities in image classification, reading comprehension, language understanding, and competition-level math. The benchmarks, like competition-level math, were nearly at zero a few years ago. By 2024, they're well above the human capability line. New top-performing models release every few months, each one erasing the records of the last.

The Human Side: A New Twist

Organizationally, we aren't pulling our weight. Only 11% of HR/Learning & Development leaders feel extremely confident in their future skills-building strategies (2026 L&D industry report). A January 2026 Fortune/ManpowerGroup report finds that the more workers use AI, the less they trust it. Exposure without structure breeds skepticism, not skill.

“The gap is not the technology, but it’s more the lack of tools and training that’s driving some of this anxiety,” Mara Stefan, VP of global insights at ManpowerGroup

Framing through 5T Thinking: we're investing in Technology while under-funding Technique and not creating the Times to learn by doing. IDC estimates that skills shortages could cost the global economy $5.5 trillion by 2026.

What to Do

If you're in leadership, give the Humans-in-the-Loop more than access. BCG's 2025 AI at Work report is useful here in highlighting that high-performing organizations take structured time to experiment and use feedback loops to build judgment, not just usage. The report talks about “reshaping workflows,” what is more generally called crafting your work.

If you're an individual contributor: don't wait for training. Spend a few minutes every day running a small experiment. Ask a chatbot to draft a section of a report, critique your work, or prep you for a hard meeting. The goal is to develop judgment about when AI helps, not just using it more. Did I have to fact-check everything in the draft for this post? Yes. Did the AI get some of the graphical interpretation wrong, yes. But stop for a second… Claude Cowork looked at a cut-and-pasted graphic and offered an interpretation. I’m still impressed, and it certainly types faster than I do.

Small experiments, done intentionally, compound. The technology upgrades daily. How will you?

Disclosures: I leverage every AI tool I can get my hands on as I write these posts, many of them on paid accounts and some with persistent memory.

I’d waited long enough. I've been watching Alexandra Samuel develop her persistent and agentic AI assistant, Viv, since the summer of 2024. (If you haven't listened to the TVO podcast "Me Plus Viv," it's a fascinating and musical window into what a human-AI working relationship can be.) Yesterday, I built my persistent AI assistant. Yes, I picked a cool name, and no, it’s not HAL.

What Is a Persistent AI Assistant?

No, I don’t mean an assistant that bugs you all the time. AI chatbots like Claude, ChatGPT, or Gemini don't remember anything from one session to the next. A persistent AI assistant solves this by storing structured text files in a folder on your computer that the AI reads at the start of each session and updates at the end. Think of it like hospital shift-change notes: the outgoing nurse leaves a briefing for the incoming one so patient care is continuous.

The folder becomes the AI's memory. It contains an instruction file that defines who the assistant is and how it should behave, memory files that capture your profile and preferences, handoff notes from previous sessions, and a simple thread tracker to prevent ongoing projects from being dropped. No coding required. Everything is plain text files that you can read and edit yourself. I’ve created separate GPTs for my writing style and course design, but those are black boxes.

The Persistent AI Assistant How-To

I started from Alexandra Samuel's work: her HBR article on building your own AI assistant and her piece on three structures for your AI team. Once I had my own persistent AI assistant up and running, I wanted to better understand what I’d done. For example, how the files were structured, whether the assistant would run into context window limits, what the design decisions were, and whether someone without a technical background could replicate the process. The document here is the result.

I gave this prompt to my infant/journeyman assistant: "We have done a lot of work to create the agent. Please create a 'human-readable' document that would let a non-tech savvy person do the same thing within their Claude instance." The Google doc how-to is about 80% from my assistant, 10% my suggestions via additional prompts, and 10% my direct edits.

Why Bother?

The difference between a generic AI chat and a persistent assistant is the difference between talking to a stranger every day and working with a colleague who knows your style. My assistant knows I write in first person, cite in APA 7.0, care about the distinction between augmentation and replacement as people learn to work with AI, and that when I say "the Annals" I mean the Academy of Management Annals. It knows my active research projects, my co-authors, and my frameworks. I don't re-explain any of this. I can just get started.

That kind of continuity changes what you can do with AI. Instead of spending the first five minutes of every session re-establishing context, you can pick up a thread from last Tuesday, ask for a revision to a draft your assistant has already seen, or say "use the 5Ts" and get output grounded in your own intellectual framework (and less likely a hallucinated version if you emphasize the importance of being factual and providing references).

A Hint

This is not magic, nor effortless -- but it was fun. My process was iterative, and it involved conversations in which I taught the assistant about my work by feeding it my papers, blog posts, and chat histories. The guide makes it look clean and sequential because that's what a guide should do, but the actual experience was more like training a new research assistant: a mix of explicit instruction, correction, and learning what works by trying things. The best part: I don’t feel guilty if I walk away.

It also requires a specific platform. The guide is written for Claude Desktop's Cowork mode, which gives the AI access to a folder on your computer. If you're using ChatGPT, Gemini, or another tool, the underlying principles still apply: persistent context through structured files, but the mechanics will differ.

The hint: See what happens if you just tell Claude to use the Google doc to build your own assistant. Maybe my next step will be to have my assistant build an AI-readable document rather than this one for humans.

Sharing Flywheel

Creating a sharing flywheel is a concept I share in some of my AI workshops. We all learn more the more we share. If you build your own persistent assistant, I'd genuinely appreciate hearing how it goes. Check out the comment button below.

Disclosures: I started this process with a paid Claude Pro account. I eventually maxed out my daily limit and jumped into a Max account for this month. If I had more patience, I could have stayed on my Pro subscription. The document shared here was drafted by my persistent AI assistant following my editorial guidance, direct edits, and contributions from ever-present Grammarly. The process described in the guide is based on my personal experience and is not affiliated with or endorsed by Anthropic.

You’ve honed your AI skills. You may even subscribe to multiple AI platforms to access the best version of whichever generative AI is hot today. You’ve used AI to train for your career and figure out which industry will give you the best opportunity for growth. Now it’s time to apply for jobs. Do you go cold turkey and ghost your AI? 

Making Rules on the Fly

No one says using AI to find job openings is a bad idea. However, plenty of jobs require you to certify that you haven’t used generative AI to produce the application. Today’s job application process is an ethical tug-of-war with real-world consequences.

Is the organization hiring you or you and your tools? Chefs bring their own knives to work. Do you show up with your tools or leave your trained AI at the door? 

Do the expectations change across spellcheck, text auto-complete, Grammarly offering you a sentence with greater “clarity,” ChatGPT outlining, editing, crafting the complete document, or leveraging a purpose-built AI-job application platform that submits applications on your behalf? (Hint: Blanketing your field with applications may not be the best idea.)

Ironically, Anthropic, the company behind the Claude AI, includes this instruction in its job applications: 

“While we encourage people to use AI systems during their role to help them work faster and more effectively, please do not use AI assistants during the application process. We want to understand your personal interest in Anthropic without mediation through an AI system, and we also want to evaluate your non-AI-assisted communication skills.”

No Easy Answers

Our research suggests many people think about AI (and other technology tools) as a silver bullet or a poison pill: a unidimensional perspective that a particular technology (like AI) is good or bad. However, leveraging AI in your job application (or at work) is likely not a yes/no situation.

Many organizations have complex and shifting preferences about AI use. Using generative AI instead of traditional email templates may be a yes, while image-creator AI may be a no. However, if the organization’s broader platforms (like Microsoft 365) have embedded AI, the organization may encourage art generation through that platform. What’s policy today is likely not the policy of tomorrow -- compare this to the work-from-home/return-to-work merry-go-round.

Nuance, Not How-Tos

There are no unidimensional answers about whether or how to apply AI in the job application process. Instead, the outcome results from a particular combination of “5Ts.” Consider your talent (e.g., skills), technology (e.g., AI strengths and weaknesses), and technique (e.g., the prompts you use and the rules stated in the job application). The quality of the combination comes from how well you align the mix with your target (goal) and your times (context). 

Your target is to land the job that best suits your interests, skills, and preferences. We live in a time of wide-ranging and rapidly changing tools and expectations about AI use. Take stock of your talent, technology, and technique.

Talent

Has your work with AI given you the expertise to work without AI? Working with Grammarly has me on high alert for “passive voice” and unclear pronouns in my writing. That said, take care that you aren’t learning bad habits from biased AI or giving AI outputs more credence than deserved. Use AI to augment your learning, not avoid it.

Technology

Did you train your AI on your own work (I have a chatbot I trained on my academic publications that helps me edit new work), or did you leverage an AI trained on other people’s skills? Do you know the particular risks and benefits of the AI you use?

Technique

Did the AI do the work or help you do the work? Grammarly’s new Authorship tool can help us track where information came from. However, the Grammarly tool may also be an escalation of organizational tracking. 

Are you ready to describe how you used technology to create your work sample or application? Does your perception of authorship match the reality of how you used the AI?

Challenge or Opportunity

Set aside the ridiculous, though possible, spiral where an AI screens your AI-generated materials, you get the interview, then leverage an AI video filter to enhance your appearance (don’t we all?) and AI speech recognition to suggest possible answers as you interview with the organization’s AI avatar. 

Instead, see your decisions about using AI in your application as preparation for your interview. Prepare to discuss the nuances of modern work design with the interviewer. Clarify how much discretion you will have with AI options and practices in your work. Talk about whether in-house and external training are available to maximize your human-in-the-loop opportunities. Be more than the best human or AI for the job.

• Your organization needs to spend more on supporting the “Humans-In-The-Loop.”

• You need to spend more time supporting your own application of these tools to your work.

Let’s start with the pace of generative AI. Anthropic offered the AI data above in June of 2024. The chart plots benchmark capabilities of leading generative models from OpenAI, Anthropic, and Google. When I share this data in class, I show a ChatGPT-created image of the different genAI platforms as frogs doing what frogs do, leapfrogging (join me in our new AI Empowered Leadership program and you can see this fine image.)

I’ve augmented the original chart with Wharton and Slack's human usage and comfort data, respectively. 

How About the Humans?

Wharton’s 2024 AI Adoption Report offers a stark contrast when you compare support for the Humans-In-The-Loop to the sharp upward trend of genAI capabilities and use. “[J]ust over half of companies currently provide courses on AI, hands-on access, and internal projects or pilots” (p. 47). Most organizations offer tools, but few provide structured opportunities to reflect on how to use them in real work. 

It’s like handing out electric bikes but no helmets, paths, or charging stations. (Cheers to ChatGPT 4o for this simile)

Slack’s Fall 2024 Workforce Index adds another angle. Their survey of global knowledge workers finds that while 99% of executives spend money on AI, almost half of desk workers are uncomfortable admitting to management that they use AI. Framing this in 5T Thinking, we’re offering the technology but not the technique — or the slack time — to integrate AI in a way that works for the talent.

We Can Keep Up

If you're in leadership, support the humans-in-the-loop across the 5Ts of Talent, Technology, Technique, all aligned to a Target and our Times. 

If you’re an individual contributor, stop waiting for perfect training. Spend a few minutes everyday trying something new: Ask a chatbot to rewrite a report summary, analyze a dataset, help prep for a meeting, or create an image for your next presentation. You’ll get the time back as you learn to augment your work.

Small experiments lead to capability gains — even if the first few are flops. If you tried something last week, you may get a different result this week or with a different tool.

To be sure, AI isn’t Robin to your Batman. It’s a quickly evolving toolset. Our human performance won’t upgrade overnight, but like savings interest, our capabilities can compound — if we treat ourselves and our teams as learners, not just users. The big genAI platforms are spending millions on training the AI. Spend a little time on yourself.

So yes, AI upgrades daily. But the more interesting question is: how will you?

Spoiler alert: Yes, we are.

Amy Yee of the Wired for Change podcast invited me to my neighborhood studio (until then, I didn’t know I had a neighborhood studio!) for a fantastic conversation about the Innovation Skills Gap We Can't Afford to Ignore. Amy found a way to distill decades of my research and teaching into 45 crystal clear minutes. 

Whether this is a review or your first time hearing about Thinking in 5T; systems savvy; or the three-step loop of stop-look-listen, mix, and share—we’re sure you’ll find insights to use this week, at home and at work.

Thank you, Amy. I’m looking forward to learning more from the Wired for Change podcast.

AI can give you and your employees superpowers. However, AI is not (at least yet) a teammate. Set your expectations realistically, keep your options open, and think of AI as pieces of Iron Man’s armor or the tools on Batman’s utility belt. AI is not your sidekick, but it can be fantastic. 

Pitfalls of the Teammate Mindset 

Thinking of AI as a teammate is misleading and can foster unrealistic expectations. Any time we get the negative surprise that something, or someone, isn’t working with us toward the same goal, it’s disappointing. We grow up looking to be picked for a team. We look for synergies (and lunch companions) when we build a team at work. We see office wall posters noting there is no “I” in team. These team expectations focus on coming together to work toward a goal. AI can help us reach our goals, but it’s not yet ready for the “coming together” part. We don’t have joint accountability with AI. We choose to work with AI, the AI don’t choose us.  

AI as a Collection of Superpowers 

Instead of thinking of an AI as a teammate, see it as a suite of tools that give you various superpowers. For example, ChatGPT, CoPilot, and Claude can use natural language processing (NLP) algorithms to summarize reports or follow your request to adjust the wording of an email. DALL-E can create images like the one above. CoPilot within Excel can help you visualize your data. Many different tools can help you summarize calls with your human team members and even generate to-dos from the meeting notes. 

If you have a task to do, there is likely an AI that can supercharge your approach. Still, we are in the early stages of staffing our work with AI. Andrew Rabinovich, vice president and head of AI and machine learning at Upwork says, “In the future, you’ll just have one window on your screen where your AI companion — every person is going to have their own — is going to be hyper-personalized. It’ll understand everything that you do.”  https://sloanreview.mit.edu/audio/never-too-much-ai-upworks-andrew-rabinovich/  

But we’re not there yet.  

Harnessing AI's Superpowers 

To harness the power of AI for yourself or those you’re helping in your organization, adopt a mindset of augmentation rather than partnership. (Here are some steps from a prior post.) Think about your work in terms of tasks. Engage in an AI chat where you ask the AI how it might help you. Glance through websites cataloging different AI and their capabilities. 

Break the work down such that the AI does what it’s good at, and you keep those things where you’re better than any tool. Medical imaging has used AI for diagnosis for a while, but they are very careful to know the machine's capabilities and whether the people using the tools are novices or experts.  

Focusing on the tools we can all access for free or nominal charges right now, AI-driven automation handles repetitive and mundane tasks, freeing up your time for more creative and strategic pieces of the work. By thinking of AI as your supersuit, you amplify your productivity and innovation, achieving outcomes that you might not be able to do without them. Ironman can’t fly without his suit. Batman is just Bruce Wayne without his utility belt. 

The Importance of a Human-Centric Approach 

Many corporate AI designers and academic AI researchers focus on a human-centric approach. This involves acknowledging the unique strengths and limitations of both humans and AI and discovering ways to integrate them. Recall from my posts on Thinking in 5T: you can’t change just one thing. Leverage your skills, AI capabilities, and the organization’s processes and ability to get the value of AI in your work. Maintain a focus on the quality of the data you use with AI, your data security, and how you craft your work. 

Superpowers, yes. Teammate, no. AI can revolutionize your work. But remember that the true power lies in YOU. 

Zapier’s self-description is: “Zapier is a no-code automation tool that lets you connect your apps into automated workflows.” (See my earlier post on no-code/low-code, or check out Phil Simon’s book, Low-Code/No-Code: Citizen Developers and the Surprising Future of Business Applications.) 

Currently, my most used “Zaps” help me track my time in Canada -- information required for my Canadian immigration status. My “Into Canada,” “Leave Canada” zaps are triggered by tapping an Apple Shortcut on my phone that connects to Zapier, which then appends the date and time into a Google Spreadsheet. This automated workflow saves me from manually tracking dates and times and then having to find time to enter that data so it stays current and accurate. More work-oriented zaps focus on project management and note-taking.

Automating Pieces of Your Workflow

Wherever we have a workflow — some recipe for how we do our work — we can likely leverage automation tools to help us increase our productivity. The trick is to have a process that gives us time to think about what to automate. Certainly, we must consider foundational requirements around data quality, strong security, and work-crafting skills. But the hardest part for me is finding a moment to think about which workflows to automate rather than just diving in and doing the work. That’s where Zapier’s inspiration blog and other feeds I track are useful. Having those posts pop up is a reminder to think about my opportunities. 

High-Level Process

My overall process is the same for any process change I take on:

• Stop-Look-Listen: What task have I repeated recently that offers no or little value for a hands-on approach? (I leave some things to do by hand as that process forces me to engage with the data—generally tasks where I’m looking for insights.) What tasks do I hate doing? What are my friends an colleagues finding value from automating?

• Mix: Is my handcrafted recipe the best recipe? (Don’t just pave your workflow cow paths.) Am I thinking through my new automated mix as a negotiation? The way I craft my automations should not hinder my teammates’ processes. Reaching out to my colleagues can be a help. By collaborating and adjusting our approaches together, we all stand to gain.

• Share: I’m sharing insights here, and I'm confident many of your colleagues have strategies they’re eager to share. Your colleagues are likely interested in new approaches you’ve found. There's no reason to repeat the same mistakes and every reason to build on your colleagues’ successes. Likewise, if many of us encounter the same obstacles, it might be time to rethink our strategies.

Are the Recipes Better with AI?

Zapier and the other no-code platforms leverage AI from the top down, so we can leverage AI from the bottom up. 

That said, AI isn’t (yet) great at everything. Juliane Bergmann’s recent article in Wired offers some amusing (and time wasting) examples of AI not being a value add to her workflow. 

How do we decide? The 5Ts can help here. Start with your Target. Consider your Times (context). Consider how the particular Technology matches your Talent and Technique. …and I’ll add a new piece to the process: Set a timer for how long you’re willing to play around with your mix. This can be a simple as “ok, I’ll spend 30 minutes to see if this bot is a help,” or “I’ll set a calendar reminder to cancel this free trial account three days before the first bill will post.” Juliane’s example of falling into a $200 membership fee reminded me of when I played with Google AdWords and cost myself over $1000. Lesson learned the hard way.

Find a Flow

More broadly, remember this isn’t “one and done.” Consider these workflow automations as “minimum viable products” that you’re testing via light-weight experiments and prototypes. Don’t get committed until you know your new approach is a step in the right direction — and certainly don’t feel bad about dropping a change that isn’t working for you. Crafting our work is an on-going process.

What signals do you use to determine when it’s time to consider making an adjustment to your workflow? What timeframe works best for you? Please share below.

Disclosures: I use paid accounts with Zapier, Grammarly, OpenAI, MidJourney, Google, and Microsoft Copilot.

Work crafting is when workers independently (not guided by management) adjust their work to better suit their skills, preferences, and performance. The academic literature typically calls this “job crafting.” Still, I use “work crafting” as that term keeps us focused on the idea that anyone can do this whether or not they are in a traditional job setting.  

A business leader asked me last week what foundational skills their employees need to effectively apply AI from the bottom up. (The organization already had a good handle on top-down applications of AI in its business practices.) 

My answer to him is generalizable to us all.

Foundational Building Blocks

• Good data

• Good security

• Crafting skills

Before effectively leveraging AI to enhance and redefine our work processes through bottom-up work crafting, we must ensure that we have a scaffold of data security built on a foundation of accurate and well-understood data, energized by a culture of experimentation and work crafting.

This holistic approach is not merely preparatory but essential, setting the stage for a future where AI capabilities and human ingenuity converge as we create innovative and resilient work organizations. Many of you may see the opportunity to Think in 5T—and you’re right!

Our Target revolves around effective performance, the security of our organization and clients’ data, and the quality of our work lives

The Times are, as so often noted, unprecedented in the pace and capabilities of AI change

Talent means all of us in this context. We all need to know our options regarding data and integrations we can leverage, the tools we can bring to bear, and the security boundaries we cannot cross. In this case, there is a tight link between Talent and Technology. This interplay between Talent and Technology is critical, as it shapes our approach to leveraging AI and other digital tools.

The Technology node includes our security technologies, “white-listed” AI, and all the other tools we’re considering.

Technique is how we bring our talent and technology together. Performance management that supports the processing and training across security, available data and integrations, work crafting and bottom-up AI applications are effective techniques to the extent that we manage across the 5Ts in concert.

My colleagues and I have a well-known paper entitled, Information Technology and the Changing Fabric of Organization. If I had it to do over, I would argue for something more rigid than “fabric.” Security and data quality are more immutable than fabric suggests.

The good news is that training and resources around security and available data and integrations can be relatively stable. Similarly, the motivation and practices of work crafting are well known. Finding the best tools for the job is where it’s hard to keep up. 

Foundational Practices

Prepare your organization to leverage AI from the bottom-up. Evidence of the benefits (and complexities of implementation) means we all need to pay attention.

Stop-Look-Listen -- both internally and externally. Does everyone know the kind of data available to support their work, how to access it, and what integrations are feasible? Deeper understanding can be beneficial as people understand why data quality is critical in all our work.

Mix -- negotiate changes such that different stakeholders and requirements are involved. My work crafting shouldn’t be detrimental to my teammates’ process. We can benefit by working together as we make our adjustments.

Share -- I’m sharing here, and I bet many of your colleagues also have strategies they’d be happy to share. Develop a routine that helps you stay aware (internally and externally) of practices that work and those that are off the mark. No need to keep making the same mistakes. Similarly, if many of us are hitting the same roadblock, it may be time for some reengineering.

In the Northern Hemisphere, it’s that time of the year. First day of school pictures, backpacks full of books versus hiking and beach gear, and a flurry of back-to-school sales. Fall of 2023 is also the first year we have a bit of an organized response to the availability of generative AI -- AI trained on massive datasets of prior work and information that then can create new work when we “prompt” it. While you might get away with last year’s outfits this Fall, you cannot ignore how AI is changing education.

Spring 2023: Students 1, Teachers ??

As schools let out last spring, many students had more AI expertise than their teachers. Not surprising as one of the easiest things to do with generative AIs like ChatGPT, Bing, or Claude 2 is to generate text in response to prompts. Much homework in the past was a prompt (to a student) to generate text. Easy 1:1 swap of effort from person to AI. 

Schools and teachers hadn’t had time to prepare new assignments. Teachers and parents had little time to think about updating expectations and values around what is and isn’t cheating. None of us, even today, have had much time to consider the huge shift in what work is most valuable and important for people to do versus outsourcing to tools. (I argue for “owning” your own tools.)

Some of these questions were first asked as Wikipedia… or take a step back, calculators, came on the scene. But this year, it’s different. Generative AI can do much of our (and our students’) work.

What We Need to Think About - And Where to Take Action

Teachers open class this Fall after a summer of whirlwind catch-up. How do we support students' development in a world filled with an exploding number of AI tools? I’ve never heard a teacher asking for spell-checking to be turned off, but whole school districts have banned (tried to ban?) AI. 

High-Level Evaluations

AI in education is a global topic. I’ve scanned a variety of reports (UNESCO’s report is 418 pages, scanning is the best I can do) and will share a few insights here. The May 2023 US Office of Education’s Artificial Intelligence and the Future of Teaching and Learning: Insights and Recommendations offers much to cherry-pick:

Some headings:

• A Duality: Learning With and About AI

• A Challenge: Systems Thinking About AI in Education

• Preparing and Supporting Teachers in Planning and Reflecting

These resonate with me, given my focus on Thinking in 5T as we build any kind of human, social, or technological system.

Specifics

The report also suggests possible benefits from leveraging AI in education: “Personalized learning, project-based learning, learning from visualizations, simulations, and virtual reality, as well as learning across school, community, and familial settings.” These are generally approaches we know are valuable, but given limited resources, hard to do. AI may change some of this calculus.

When asked how my courses are changing, I highlight that I can ask the students (all working professionals) to do more as they apply our course concepts to their work settings. I can also expect higher quality. I describe the value of AI as offering them an intern. Yes, having an intern has costs and benefits, but you both get an extra set of “hands” and a chance to be in a teaching role. Both are great opportunities for performance and deeper learning.

The 2023 US Department of Education report draws on their 2017 report (AI mentioned once then) to offer learning goals and considerations on how technology can be supportive. They say:

1. Enabling Enhanced Question Types: to give students more ways to show what they know and can do.

2. Measurement of Complex Competencies: to better elicit growth in important skills that go beyond typical subject matter standards, for example, in measuring practices, social skills like teamwork, self-regulation, and work-relevant skills (e.g., making presentations or leading teams).

3. Providing Real-Time Feedback: to maintain and increase student engagement and to support effective learning, providing timely and helpful responses and suggestions to each learner.

4. Increasing Accessibility: to include neurodiverse learners and to engage learners’ best communication capabilities as they share what they know and can do.

5. Adapting to Learner Ability and Knowledge: to make assessments more precise and efficient.

6. Embedded Assessment in the Learning Process: to emphasize an assessment’s role in improving teaching and learning

7. Assess for Ongoing Learning: to reveal progress over time and not just predetermined milestones.”

2023-2024: Our First Year of Experimentation

Governments and non-governmental organizations took until the summer to get us to their assessment and recommendations. We should see this year as one of experimentation. AI offers us new tools and questions in a way that changes -- and will keep changing -- daily. 

How to Help

If you’re reading this, you are in one of my top two categories -- I’m not worried about you:

1. People who spent the summer trying out new tools and thinking about how to get value from them

2. People who were listening when others talked about their AI experiments

3. People who think they can ignore the latest batch of AI like they did blockchain and TikTok -- these are the folks we all need to be supporting

It’s a Process

As with any change, we need to stop-look-listen, and mix together a version of education and work where our talent, technology, and technique fit our targets and our times (Think in 5T). Then we need to share the results with others. As you see helpful ideas and examples, share them with your network. That’s the easy one. Include people with more and less technology experience and understanding. 

Talk with kids, teachers, and school administrators, but most importantly, figure out how to use AI in your own work. AI is too important a change to get behind. Share your understanding and ideas with your children, family, friends, and colleagues. 

If You Need Help to Get Started

Follow Professor Ethan Mollick:

• The Homework Apocalypse

• Assigning AI: Seven Ways of Using AI in Class

• Using AI to Implement Effective Teaching Strategies in Classrooms: Five Strategies, Including Prompts

Tryout Some of My Strategies

• Show and share your work process. See an example here.

• Use one coffee break a week to look back over the available tools. They change daily. Did you know you can use Microsoft’s Bing bot in Chrome? I didn’t until today.

1984-2024

Welcome to a school year to remember. I know personal computers were available before the 1984 Apple Mac Commerical, but that’s the date I use to think about computers being in wide public use. Here we are 40 years later. How will we look back on this year? 

For example, I went through IBM’s Augmented work for an automated, AI-driven world report today. The authors, Jill Goldstein, Bill Lobig, Cathy Fillare, and Christopher Nowak note, “A recent IBM Institute for Business Value (IBM IBV) survey found that 4 in 5 executives say generative AI will change employee roles and skills. But only 28% of CEOs in our 2023 CEO Study have assessed the potential impact of generative AI on their current workforce” [emphasis added].

In March, a colleague asked me if AI would be another case of the rich getting richer. I came to see her point, but now I’m equally worried that it’s a situation where not being early in adoption with AI is a bigger risk than with other new technologies. People who have been successful as late adopters may not fare as well with a wait-and-see strategy regarding AI and their jobs or AI and their business models. 

As an educator, it’s my job to help people decide what to pay attention to. You may also be inclined to help others make these assessments.

Two Calls to Action

1. As you see helpful ideas and examples, share them with your network. That’s the easy one. Your ecosystem is likely already sharing and using one another as backboards for your thinking. 

2. More importantly: Share with at least one person you believe is a step behind your own thinking. This could be a friend, a relative with a small business, or someone with a successful middle or executive management role. You may have heard them acknowledge AI but say they haven’t had time to try. Or maybe they tried a tool like ChatGPT in the new year but didn’t see the connection to their work. Have a stop-look-listen AI moment with this friend, relative, or colleague. 

Resources for Your Lucky Friend and You

Overview

While the IBM report I mentioned earlier is primarily survey-based, the breadth and depth give us much to consider. (I’m always searching for assessments with objective processes and performance data -- thank you for sharing examples in the comments below.) The authors highlight three priorities for augmenting work:

– Transform traditional processes, job roles, and organizational structures to boost productivity and enable new business and operating models.

– Build human-machine partnerships that enhance value creation and employee engagement.

– Invest in technology that lets people focus on higher value tasks and drives revenue growth.

Broad-ranging reports like this IBM one can be great motivators. Then we need action. More from the IBM report:

The human-machine partnerships that will drive advantage tomorrow are being developed today. That means workers need to be willing to experiment with new approaches to understand what works—and tech-savvy enough to troubleshoot along the way.

Work Crafting

My colleagues and I have done some deep dives on how we can leverage AI as we craft/recraft our work. The quote above points to experimentation. IBM and recent books by Prof. Amy Edmondson, and Profs. Michael Luca and Max Bazerman highlight that experiments don’t always work -- failure is part of the learning. Manage the process with lightweight experiments such that the risks are appropriate for the gains. What’s important is that we look for new approaches to our work, given the pace of change across our tools and environments.

The Stop-Look-Listen Process 

Drawing from my 2012 book, The Plugged-In Manager:

• Stop to see how things are going.

• Look to see what opportunities you have. Span the 5Ts of Talent, Technology, Technique, aligned to your Target and Times. Track sites and newsletters highlighting and reviewing new tools and approaches appropriate for your work. Make a change, and…

• Listen to the result both in your area of influence and beyond. Return to the first step, Stop. 

Depending on your goals, the pace of change in your organization, and the pace of technological evolution in your work, this process could be something you do every Tuesday or every second Tuesday (for example). Given the pace of AI today, I feel like I need to do a stop-look-listen every morning.

A Bridge to Opportunities

Whether you’re championing your own projects, extending a hand to a friend, colleague, or relative, or steering your entire organization through these tumultuous times, embrace adaptability. Think of it as building bridges to keep your paths open and be willing to reinforce and reconstruct your bridges as conditions change.

Friends of mine have recently formed a new consulting venture, Pontem Advisors. They chose “Pontem” (Latin for bridge) because they assist organizations in "bridging the gap between company business understanding, goals, and behaviors….” I see the significance of their methodology across organizations at every phase of digital transformation.

We must stop-look-listen at each step and cross bridges to new opportunities. Help yourself, and a colleague, take the time to test a new direction with AI and their work.

Resources to Help You Leverage AI In Your Work

• Futurepedia: “The Largest AI Tools Directory, Updated Daily”

• FutureTools: "FutureTools Collects & Organizes All The Best AI Tools So YOU Too Can Become Superhuman!"

• One Useful Thing: Prof. Ethan Mollick’s Substack

• Dr. Alexandra Samuel’s work across LinkedIn, the Wall Street Journal, and more

• Discussions (human-to-human or online) within your organization about what your colleagues find useful.

• Suggestions from the AI — either across or focused on the specific platform. For example:

  • Asking ChatGPT: “What's a good prompt to use for MidJourney if I need an image of AI code wrapping around the world?” As I write, ChatGPT has internet access blocked, so it’s not aware of MidJourney — but it still writes a good prompt. Give it a try!

  • General searches for capabilities and finding articles offering options. For example, Discord AI Art Generator: 7 Best Servers to Use by Mercie Mwikali

Ode to a Beautiful Search Tool

Click on the image below to look at our now “old school” (but beautiful) AI search tool. Mobile users may have a better experience here. Spring 2023 was our last update, as the search engines and AI now do a better job. Our human efforts will focus on finding new ways to craft our work with these tools.

Taking a Step Back to Build a Foundation

Does having access to all these tools seem like putting the cart in front of the horse? Take a look at this earlier discussion that Utpal Mangla (IBM) and I put together. We talk about how organization leadership can bring in top-down automation, but how all of us can address bottom-up automation in our work (and lives). Much of that conversation started through our work with ISSIP (The International Society of Service Innovation Professionals). We are honored to be the current (Utpal) and past (me) presidents of that organization. Join us — for the ISSIP upcoming Discovery Summit on Tech for Good.

Automation is part of our work. The question is whether we are finding the tools to augment our abilities or waiting for our organizations to decide what to automate and who to replace. Research finds that people who craft their work are more productive and employable. In this earlier post with Brett Li of Tonkean, we talk about that research and the steps you can take to craft your work. But what tools can help us?

Bottom-Up Automation Tools

Here is the Spring 2023 quarterly update of our database of personal automation tools! Nitish Gupta and I collect tools that leverage artificial intelligence (broadly defined) to support work. You likely use some of these tools already (e.g., GMail’s autocomplete). We hope you also find many valuable new tools within our mix. There are thousands of news tools given the generative AI explosion. Here are 200+ of our curated selections. Our criteria include these tools are:

• for work (again, broadly defined) - generally, not lifestyle. That said, many will help across your work and leisure.

• that individuals or teams can integrate into their work without organizational support. (Many powerful tools for automation would connect directly into an organization’s systems - these are better taken on as top-down and bottom-up efforts.

• leveraging automation either internally (e.g., machine learning or other artificial intelligence) or as part of a process connecting to AI tools (e.g., IFTTT).

Click on the image below to have a look. Mobile users may have a better experience here.

This Quarter

We added over 20 new tools this quarter. New/split categories include: Creative Suite, Research Support, and Code Generator.

What’s Next?

We update the list quarterly. Ironically, for now, this is a hand-crafted process. As we looked at earlier lists of automations for personal use, we found many were out of date. Tools may come and go, change names, are rolled into other services, and the like. We will use automation to check our links each quarter and add new tools as we find them — or as you suggest them in the comments ⇓ below ⇓.

Realities are barging in on my euphoria about Microsoft 365 Copilot. After a day of sharing shorter and longer versions of the Microsoft announcement, I’m starting to track my questions and those I hear from others.

Business Chat works across the LLM, the Microsoft 365 apps, and your data—your calendar, emails, chats, documents, meetings, and contacts—to do things you’ve never been able to do before. You can give it natural language prompts like “tell my team how we updated the product strategy” and it will generate a status update based on the morning’s meetings, emails, and chat threads.

I have no doubt (and even have placed bets) that our work will transform in the months and years after its release. (Still months away if you’re not in their preview group of 20 companies). Already I’m dealing with frustrations as I can’t bring myself to work on a Powerpoint deck without the flashy support of Copilot.

Deeper Considerations

My colleague, Prof. Mila Lazarova, wondered if this would be another case of the rich getting richer. I waved off her concerns by saying Copilot was part of 365. Another colleague, Prof. Matt Beane had just tweeted that 345 million people use Microsoft 365. That’s a broad base. However, a text from author Phil Simon about premium features sent me back to the announcement document.

Pricing reared its ugly head.

I hope that Copilot won’t be an upcharge beyond the basic 365 pricing. Given all the enterprise language in the announcement document, I also hope that personal and family accounts get the magic too. For years, I’ve asked for my personal “graph” to be leveraged to improve my work (albeit generally in the context of Google).

My Copilot euphoria is based on the AI ability to leverage the work we’ve already done (our graph) to support the work we need to do.

The power I’m seeing in the Microsoft demos would put knowledge workers without Copilot capabilities at a severe disadvantage. Mila would be right. Already I’m frustrated that only 20 organizations have access. I also expect my organization won’t be first in line for the next wave.

Another fear is that the work history that powers these benefits will be locked within organizational walls. Remember when you couldn’t share collaborative documents or discussions outside your organization’s domain? Yes, security and provisioning will be of huge import - but perhaps a virtuous circle of AI will make it easier to build the walls and bridges we need.

We need to own our own tools.

My biggest fear is about my personal work. Will it be locked inside my organization’s account?

The better our “graph,” the better our work. Own your own tools.

Learn more here: Microsoft 365 Copilot Site

2008

[Appologies for broken links. No copilot to track them down....] Three times in the last week I’ve heard/read about employees owning their own computers – rather than having the company provide them. This harkens back to a world where the identity of a craftsperson was partially determined by his/her tools.

First time was a friend who was switching jobs and going through the start-up costs of a new computer. Decision was finally just to turn in the company-provided laptop and use a personal one. My friend is a world-traveling consultant who needs his personal applications on the road. (Even when companies will allow you to put personal applications on a company machine, problems are created when tech support re-images the drive to do updates and fixes.  You then get to spend hours reinstalling your personal applications.)  Carrying two laptops isn’t an option.  Agreement is that the company will support its applications, owner will support his own.

Second time was on the TWIT podcast (about 54 minutes in) where the panel noted that it’s common for people to have preferences about their tools – e.g., Windows versus Mac, and/or particular applications.  John C. Dvorak noted that when he was a kid working in factories, the master mechanics owned their own tools.  The benefit the TWIT panel ascribed to the firm is that the employer doesn’t have to take on the burden of computer “refreshes” etc., and the employee gets the benefit of controlling their own applications and methods of work.  The conversation then hit on the main reason big enterprise isn’t excited about this: security.  Quick transition to the variety of technology approaches available for solving the security problem.  It can be solved.  Issue is creating safe “virtual machines” that connect to the company network.

Third time was an email summary of the week’s tech/work news from a professional group I belong to.  They linked to Rasha Madkour’s article “B.Y.O.C. (computer) to work - wave of future?”  Citrix (the company highlighted in the article) even provides a $2100 stipend to help employees buy a laptop and support package.  Key to note is that Citrix sells one of the tools that handles the virtualization issues (VMware  also provides a desktop virtualization product). 

Is the third time the charm?  Have we circled back to an age where we will again own the means of our production?

Other organizations that seem to be getting on the bandwagon:

• Microsoft & British Petrolium (see Ina Fied’s Beyond Binary CNET blog)
• Arizona State University (see article by Jennifer Mears of Network World)

How does this change our relationship with our employers?  Do we then become more valuable the better our technology? Do we become more effective given greater control over our tools? Tool ownership might tip the balance for some regarding employee versus contractor status.  (See some checklists here and here.) Note that the desktop isn’t the only tool in the box.  Access to organizational resources stored either on company servers or on company portions of the “cloud” may be where the real work gets done.

Show Your Work

Sharing the provenance of our work adds value at the moment and as you return later to leverage the ideas or make them better. However, we know from Sidney Harris’ famous cartoon of scientists working on a blackboard (“then a miracle occurs”) to math assignments worldwide, showing your work is complicated. 

Hyperlinks give us the ability to share our sources without much distraction. Various in-line academic referencing techniques (APA, MLA, etc.) offer value to insiders as we can translate “DeSanctis & Poole (1994)” in our heads and place the related material in context with decades of research. Footnotes are slightly less distracting, and endnotes push the connections out of sight. 

Yet, even traditional citations don’t tell us how we made the connections. Formal research writing is such that you should be able to replicate the work. Even so, we are in the midst of a replication crisis. Some of that is due to sketchy research. “Lucky” results are another issue. But some of the crisis must be due to limited documentation of the process. When we run an experiment, we don’t always fully describe the room setup, the software versions, or the particular timing -- we used to be better at this.

Space Constraints (Really, Attention Constraints)

I end my January piece in The Conversation: Why using AI tools like ChatGPT in my MBA innovation course is expected and not cheating with "I like an approach some photographers use: share your tools and critical settings." Susannah Schmidt, my editor on that piece, suggested I take the idea further and fully document the approach I used to create the article.

Sharing the process even just as bullet points turned out to be about one-third again the length of the article. No-go, given the word count limit for The Conversation, but here I have all the space in the world. I also respect your time, so I haven’t gone completely overboard in the notes below.

Show Your Work To Add Value

Generative AI and other AI-enabled tools can be of great help. We are all experimenting in these early days of publicly available AI. AI learn, change, and our practices continue to evolve. I expect many of us to reinvent our wheels. 

Yet, learning how to use these tools is difficult without watching someone work through the process. Observable innovations have a greater chance of adding value, but knowledge workers work on their screens, not in public. We must make our experiments visible to learn from our new workflows. How to share the process helpfully for our audiences needs examination. 

A 5T Analysis

When I talk about Thinking in 5T, I think of the interplay across Talent, Technology, and Technique as we work toward a Target and stay aligned with our Times. 

Talent

In the case of AI use for knowledge workers, talent can start with the person experimenting with AI in their work and wanting to share that workflow with others. The first step is to acknowledge that humans are easily distracted. Tracking our process takes us out of our flow. Find a way to document your work automatically so you can lean toward “freewriting.” If you’re a coder, you may have already found a way to have AI create comments in your work (I hesitate to add a link as the best tools will change -- search for something like: “ai adding comments to code.” You likely automated your reference management years back if you are a researcher. Google, Microsoft 365, etc. automate our version control, but not yet in a way we can learn from. (Am I more productive when I outline first? Maybe Microsoft Viva Insights will someday help me know.)

For the talent trying to learn a new workflow, is a video screen capture the way, or is a textual recipe better? I’ll append my recipe for this piece below.

Technology: When an AI Generates the Work

Unfortunately, the major style-guides haven’t caught up to the AI. Some institutions offer stop-gap advice -- Thank you, Seneca and Guelph-Humber. 

The tricky part is that what the AI tells me, it may not tell you… and it may not tell me the same thing twice. Style-guide-wise, this is a bit like referencing personal communications where we can’t explicitly show you our data as it’s private or ethereal. 

Technique: Current Recipe for My Written Work:

• Throwing ideas and references/links into a Google Doc with a draft title

• Both Google autocomplete and paid Grammarly run in the background (just now, I hit the tab key and back turned into background). 

• When the idea gels, rough out the outline. Outlining is iterative as I self-edit (I’m not very good at freewriting), reminding myself about the goal of the post. Create other idea files to save trimmed ideas.

• When stuck on finding the right word, cut the paragraph and paste it into ChatGPT, YouChat, or the paid version of WriteSonic. My attempts at just putting an idea into WriteSonic or Tome have yet to pay off with a good draft. Given its access to all my background work, I expect Microsoft's Copilot will do a better job. Please let the months before my access to Copilot pass quickly….

• I wait until I have a full draft before clicking the Grammarly icon and doing the full proofing pass. (Small concession to the ideas of freewriting.

• Old world -- at the end, I add in full references via Zotero or Paper.li. In the new world, I’m hoping my copilot will suggest the best references as I type… and then, ultimately, write the review section and keep it up-to-date.

• Depending on the final publication venue, I’ll cut and paste the material into its new home. For example, this post went from Google Docs into Squarespace — where Grammarly offered still more (and sometimes, different) suggestions.

The Mix

Each of the talent, technology, and technique dimensions is in flux as I do my work. My understanding of each, and the target and the times, shifts as I go. How we reference our AI support (at least until we have “prompt history”) remains a question. For a recent short article, I happily added a sentence from ChatGPT-4’s review of my draft. Going back to T is for Talent, I wouldn’t cite an editor given today’s norms. I didn’t cite that sentence. I would like “version control” to track and help me improve my work through analytics. Microsoft Viva Insights -- are you listening? Or maybe I should say, Copilot?

[I know I can do a better job on the above paragraph, but I have to move on to another task. Enter ChatGPT-4 (though, why does it so love to say, “In conclusion”?) My prompt: Rewrite this paragraph as a powerful ending to a professional blog post. Other than “In conclusion,” and the florid “the ever-changing landscape,” not bad]:

In conclusion, the ever-changing landscape of talent, technology, and technique continues to reshape the way we approach our work. As our understanding evolves, so does our approach to leveraging AI assistance in our creative process. The implementation of "prompt history" could provide valuable insight and streamline our work with AI like ChatGPT-4. Drawing from the lessons in "T is for Talent," we recognize the importance of adapting to modern norms and finding ways to incorporate AI-generated content seamlessly. The future of work could greatly benefit from enhanced "version control" systems that monitor and improve our output. Microsoft Viva Insights and Copilot, the call is clear – we are ready to embrace the evolution and revolutionize our professional landscape.

Same original paragraph, explicit request for first-person active voice:

As I navigate the dynamic realms of talent, technology, and technique in my work, I continually adapt my understanding to keep pace with ever-changing targets and circumstances. The integration of AI support, such as ChatGPT-4, into our workflow poses intriguing questions about referencing and acknowledgment. Reflecting on "T is for Talent," I realize that today's norms may not always align with traditional citation practices. I eagerly embrace the potential for advanced "version control" systems to monitor and enhance my work. Microsoft Viva Insights and Copilot, I implore you to join me in this quest for progress as we strive to revolutionize the professional world together.

Not sure I’ve ever “eagerly embraced,” nor “implored,” but again, not bad.

The talent around us benefits our work. Publication style guides call for us to cite articles, books, and personal communications, but not our editors. What does this convention cost us? Will new technologies like blockchain enable us to gain the value of better citing all our sources?

T Is for Talent -- Often Hidden

Susannah Schmidt, the editor for my recent piece in The Conversation: Why using AI tools like ChatGPT in my MBA innovation course is expected and not cheating is an excellent example of uncredited value. Beyond her knowledge of the audience and style of The Conversation, Susannah offered insights and ideas for improvement. There is irony in this given that one of my main points in the article is that using technology isn't cheating, but not citing your sources is.

I’ve had many fantastic editors over the years. We sometimes footnote an idea or analysis suggested by these collaborators, but not often. When we don't fully source our writing, we limit the provenance of our work.

Knowledge Management

I motivate my students to use citations by making connections to the value of knowledge management. A 2017 ICD survey found people spend 19% of data activities time searching for the data. How can you validate your claims and check your results if you can't recall the source of your inspiration, data, or workflow?

Counterpoint

The publication style guides may be right. Publications credit the editors in the introductory material. This acknowledgment solves the ethical issue of editorial contribution. Editors edit, enough said. 

Limiting our references may better support the creativity of those who follow. Kane and Ransbotham find that contributors to Wikipedia articles fall off as the topic matures. If a more substantial network of references makes it appear that all the good ideas are already on the table, what happens to future insights? Are we more likely to contribute if we can stand on the shoulders of giants? Or are we less likely as attention, funding, and willingness to take risks fall off in the face of more established findings? (Thanks to colleague Ian McCarthy for a tweet sending me down this path).

The Future

I had the chance for a sit-down with Phil Simon this week. He's working on a new book, The Nine: The Tectonic Forces Reshaping the Workplace. One of his forces is the application of blockchain technologiess -- global distributed ledgers that store and track digital assets and information. Just as hyperlinks can simplify how we connect to our sources, I hope we use blockchains to map the full provenance of the ideas without reducing the article's readability.

As always, I don’t expect a singular solution. We must consider a mix of talent, technology, and technique. What’s essential in this time of dramatic technological change is that we mindfully adjust our workflows with these new technologies.

Next Up

A good editor stays in your head. In this case, I hear decades of editors coaching me to stay focused. I'll save another of Susannah’s suggestions for a follow-up post on writing methodology in the age of generative AI: T is for Technology and Technique. Please let me know your thoughts and critiques in the comments below -- a direct way to document the contribution of your ideas!

ChatGPT launched on wednesday. today it crossed 1 million users!

Every good blog post needs a hook. For today, I wanted to leverage the continued growth of ChatGPT registered users. Given it hit 1 million in 5 days, where must it be now? Likely some big, juicy number that would be great for this post. I asked Chatsonic “where can I verify the number of registered users chatgpt has today?”, but it told me yet another in a long line of fibs:

“You can verify the number of registered users on ChatGPT's website. On the homepage, you can find the "Users" section, which shows the total number of registered users on the platform. You can also find the total number of active users in this section.” […there is no Users section on either the OpenAI or ChatGPT sites.]

Is it Me?

Maybe I just don’t know how to ask the right question… but I’m likely better than many. I’m known in the family as the Queen of Search (thank you, Google Advanced Search course!) and much of my research involves me crafting search requests across a variety of platforms. It may also help that searches early in my schooling cost actual money and I had to make an appointment with the university librarian to submit my request (1980s Dialog searches for academic articles).

We’re All (including the AIs) Learning as We Go

The requests I’m running these days cross ChatGPT, ChatSonic, MidJourney, DALL-E 2, and more. Each of these platforms has a slightly different set of parameters you can use. How about this one for MidJourney: “a seamless tileable jade tree pattern, spiral carvings, octane renderer, trending on CGsociety --ar 1:1 --q 2”. Not a phrase I would have come up with, but the result is lovely. There is also the complexity of integrating some of these capabilities into our standard tools (e.g., ChatSonic extension for Chrome).

Thank you to those who have gone before and their willingness to write instruction guides -- see my short list below. Thank you too, to the platforms for offering up backgrounders to help us learn the nuances and see more opportunities. I’ll list some of those below as well.

What is the Expertise We Need?

Queen of Search may be a great title, but I wonder if we will find it on many LinkedIn or Indeed job posts. Prompt Engineer is an actual job title, but I’d like to see it fade away. Two reasons: First, I’d like all my engineers to be prompt, and second, given we are unlikely ever fully to understand the AI black boxes, is it engineering? 

How About AI Wrangler?

AI Wrangler builds on the short history of Robot Wranglers. It also reminds me of the skills needed to herd literal or figurative cats. Until, if ever, we know what it means to be fluent in the language of AI, wrangling is more in line with the reality of this work than engineering. 

Learning from Others

Prof. Ethan Mollick offers examples that show me that I have much to learn as I work with these new tools:

write an essay with the following points. use an academic tone. use at least one clear example. make it concise. write for a well-informed audience. use a style like the New Yorker. make it at least 7 paragraphs. vary the language in each one. end with an ominous note. -Humans are prone to error -Most errors are not that important -In complex systems, some errors are catastrophic -Catastrophes cannot be avoided….

Sometimes the memory is useful, you can (and should) ask it to revise previous work: change the third paragraph to be more professional or use a different example in the middle and it will provide a revised version. However, sometimes it forgets what you were working on, and you have to remind it. You may, for example, want to tell it revise the third paragraph on the essay on catastrophes so it doesn’t lose track.

And It’s Not Just the AI - We Need to Work Collaboratively

Yet, focusing on prompts seems limiting. If my goal is to race with the machine, I want to see the process become more of a collaboration. Alexa and Google Home sometimes offer options after answering questions, but usually these are more of an annoyance than a help. One more thing to say “no” or “stop” to. Coaching from the new AI would help me. Live-action versions of the backgrounders would be nice. Humans learn best actively -- when we’re in the middle of a task is the best teachable moment. ChatGPT seems the closest of the tools I’ve used in that it asks for feedback on the quality of the response. Sometimes, it also gives a hint. In this screenshot, it coaches me on how to find what I’m looking for.

My Attempt to Be Proactive

I asked ChatGPT and ChatSonic to give me hashtags for my last post. “Act as a coach, how can I more effectively ask you for help in creating good hashtags for this post?”

ChatGPT’s response:

1. Be specific: Provide as much context as possible about the post, including the topic, audience, and goals.

2. Provide an idea of what you're looking for: Are you looking for hashtags that are popular, unique, or specific to a certain community?

3. Be open to feedback: Be prepared to receive suggestions and be willing to make revisions based on my feedback.

4. Give a good example of the post, if possible. Tell me if you have any specific guidelines that you need to follow, such as the number of hashtags, length, etc.

5. And finally, be clear and concise in your question. For example, "Can you help me create 5 hashtags for a post about vegan food targeting a health-conscious audience?"

Learning Together

We are writing thousands of articles in hundreds of languages about using these new tools. (WriteSonic even offers example prompts for AI-related posts in its menu bar.) I’m reading and experimenting with as much of this as I can. I hope my colleagues and students are as well. Every time I write, stumble at getting the AI to do what I want, or comment on a colleague’s related post, I’m hoping this work is training the various AI as well as giving me a chance to practice. I’m willing to put the work in now for better tools in the future. 

Take your AI wrangling on as a challenge. If you spend much time interacting with voice assistants like Alexa and Siri, or maybe your smart oven, you know that (at least today) you must adjust your language. You may adjust your pace, tone, and vocabulary to get the desired result. Working with the new batch of generative AI is the same. …and these are the same skills you may have had to develop if you’ve programmed in anything from COBOL to Python: We start with a “Hello, World” program and move on from there (…and have no worries, ChatGPT can write those “Hello, World” programs for you and explain the work -- just ask).

The good news is that, like with programming in different computer languages, once you learn one, the next is easier. Consider other life skills that you can draw on as you learn these AI languages. In a constructive Medium post, Shubham Saboo says, “In a way, prompt design is like playing a game of charades!” Unfortunately, if I’ve played charades, it’s less than three times. Good thing I have COBOL to fall back on.

Thank you for letting me share my learning journey.

Resources

Each of the following is a support document I’ve found helpful. Thank you for sharing others you find (or that you had the AI find for you) in the comments section below.

• ChatGPT:

  • https://github.com/f/awesome-chatgpt-prompts

  • https://fka.gumroad.com/l/art-of-chatgpt-prompting

• MidJourney: https://midjourney.gitbook.io/docs/resource-links/guide-to-prompting

• More Image Examples: https://mpost.io/top-50-text-to-image-prompts-for-ai-art-generators-midjourney-and-dall-e/

• History of AI Art: https://www.stevemurch.com/engines-of-wow-ai-art-comes-of-age/2022/12

• WriteSonic: https://docs.writesonic.com/docs

I finished my Fall MBA teaching before the explosion of ChatGPT - a chatbot that can write, code, and answer questions. The next time I teach this material will be in a different world. ChatGPT reached 1 million registered users in 5 days and continues to grow. 

My innovation course for our Management of Technology MBA is writing intensive. True, even this Fall, AIs were available that could paraphrase (e.g., Quillbot), but they hadn't made the mainstream in my community. Even Grammarly, a proofreading tool I pay for and encourage my students to use, wasn't common knowledge. 

Past students don't have second thoughts about using Google Autocomplete. However, when it comes to Grammarly's increasingly powerful rephrasing capabilities, some wonder whether using such tools is cheating. My syllabus has an extensive academic integrity section, and I spend much time on how to quote and cite in business writing correctly. I also let the students know I use Turnitin.com to check for citation mistakes. (I'm happy to say I haven't encountered full-blown plagiarism at my current university.) My students' concern makes sense. 

My Answer So Far

My answer has been that no, the use of an editor, whether human or machine, just makes sense -- it's not cheating. The final words must belong to the listed authors, but good business practice supports using discussion and editing tools to get you to that final stage.

Generative AI

I will teach my next course in a world of ChatGPT, WriteSonic, and the multitude of generative writing AI that are not just editors. Examples from homepages of some of the commercial tools:

• WriteSonic: "Best AI Writer for Creating" [and then an animation lists LinkedIn posts, SEO Content, Product Descriptions, etc.] for bloggers, freelancers, etc.

• Copy.ai: "Say 'goodbye' to the blank page for good."

• Sudowrite: "Sudowrite is the non-judgmental, always-there-to-read-one-more-draft, never-runs-out-of-ideas-even-at-3am, AI writing partner you always wanted."

• Notion AI: "Leverage the limitless power of AI in any Notion page. Write faster, think bigger, and augment your creativity. Like magic!"
  
  

I expect each tool to do an adequate job (low "B") with my assignments. Leveraging multiple tools to get the design, text, and references in line seems realistic. My final individual assignment is to create facilitator-focused playbooks related to three of the course concepts. I've offered a baseline quality example to the students for the last couple of years. Before the next course, I'll redo that example using a set of generative AI.

But is using an AI in such a way cheating?

Recall the bruhaha of Fall 2022 when an AI-generated image won at the Colorado State Fair. My reply to the discussion on Twitter:

Is the prize for the art, the best human art, art by someone living in that country, art by a child, or art by an elephant (or the best monkey selfie)? We've always made distinctions, now we just need to build a norm for attribution. Share the tools and who/what submitted.

— Terri Griffith (@terrigriffith) September 21, 2022

Chatsonic seems to agree with me. I asked it, "what is the biggest concern faculty have about generative ai?" Chatsonic's reply:

The biggest concern faculty have about generative AI is its potential for creating automated plagiarism and copyright infringement. Generative AI can generate content that could easily be plagiarized without proper attribution, and its models can reshape existing material into something new and unrecognizable. Therefore, it is important to ensure that any use of generative AI is properly attributed and that the content generated is unique and non-plagiarized.

Chatsonic differs from ChatGPT in access to current web search results. To verify my understanding, I asked Chatsonic, "can chatsonic use 2023 search results?" Its reply: "Yes, absolutely! I can use the latest online search results from the year 2023 to provide you with the most up-to-date information." (I've installed the Chatsonic chrome extension such that any searches I do from the Chrome toolbar that end in a ?-mark offer Chatsonic's reply and the Google search results.)

Using AI Is Not Cheating. But Not Citing Your Source Is

When I shared three of my favorite business books in my final 2022 post, I used ChatGPT to create an alternate review of Power and Prediction: The Disruptive Economics of Artificial Intelligence by Ajay Agrawal, Joshua Gans, and Avi Goldfarb. I said, "I hope knowledge workers [people using generative AI] will share their tools just like photographers often caption their images with a description of their camera and setup." Where I used ChatGPT's work, I used the quote format (as I've done here). 

I also specified the prompt I used. However, I wouldn't expect that same prompt to give me or anyone else the same answer each time. Some image-generating AI (e.g., MidJourney) let you define a seed so that you can "steady" the results. I have yet to find similar features for chatbots. If replication isn't possible, then references to AI-generated text are similar to how we reference personal communications or direct quotes from interviews -- situations where the author has access to the material, but others do not. 

We Need APA/MLA/Chicago Manual of Style Citation Guidance

I asked Chatsonic, "give me the URL showing how to reference material generated by an AI in the chicago manual of style format?" The response was, "Sure! Here is the URL to the Chicago Manual of Style guide on citing materials generated by AI: https://www.chicagomanualofstyle.org/citation-quick-guide/citation-quick-guide-3/ai-generated-material.html Hope this helps!" Following that link gave me the error noted in the image. Entering "ai generated work" into the Chicago Manual of Style search bar returned, "Your search 'ai generated work' did not match any documents.” [Spelling out artificial intelligence was also a bust.]

I'd love to know more about the underlying language model Chatsonic uses such that it seems to fake a URL. I'd be happier about paying for the service if it would at least check itself before offering up the link. ChatGPT took an honest pass:

I'm sorry, I'm not able to provide URLs. However, The Chicago Manual of Style does not have specific guidelines for citing material generated by AI. The best way to cite such material would be to consult with your instructor or editor for guidance. It may be helpful to include information about the specific AI model or program used to generate the material, as well as the date and any other relevant details. [ChatGPT shares that it's training data goes up to 2021.]

We're All Learning As We Go

I started blogging to allow myself to reflect before answering common and more detailed student and executive questions. Thank you for joining me on this journey as we all (and I include the AI) learn as we go. I'm very interested in differing perspectives. Feel free to comment below or reach out on other platforms.

Created using a combination of Grammarly (paid), Chatsonic (paid), ChatGPT, Google Autocomplete, DALL-E 2, and the Five template on Squarespace.

Happy holidays!

My comments are in the order I read the books. You can follow my Goodreads list here. So many fantastic books to read!

The No Club: Putting a Stop to Women's Dead-End Work by Linda Babcock, Brenda Peyser, Lise Vesterlund, and Laurie Weingart

The No Club was an easy purchase (and the book for which I've made the most 1:1 recommendations). Anytime I'm asked about women negotiating in the workplace, I cite Professor Linda Babcock and women and the workplace authority Sara Laschever's earlier book, Women Don't Ask: Negotiation and the Gender Divide. Here, Professors Linda Babcock, Brenda Peyser, Lise Vesterlund, and Laurie Weingart take a discussion about their workloads and turn it into a science-backed playbook for women in organizations (and life).

They start with identifying the critical issue that brought them together: Women do more non-promotable tasks than their male colleagues. "A non-promotable task matters to your organization but will not help you advance your career" (pp. 16-17). They build out strategies for assessing the situation, making effective decisions about what work to take on, and how to effectively say no.

The Basics

• Get the Information You Need to Understand the Task

• Consider Who Is Asking You

• Avoid the Traps that Lead to Yes -- don't underestimate the costs (something I do all the time)

• Use a "Yes, No, Yes" Strategy (p. 131-132). Here’s an example they share:

"Thank you for thinking of me, Joe. I have so much on my schedule the next two quarters with leading the new product launch that I won't be able to devote the time that organizing the meeting really requires, so I'm going to have to decline. But Don would be a great person to take on the task, and he could benefit from the opportunity to interact with others from across the company. I am confident he'd do a fantastic job." 

The Long Term

Having given us an individual playbook, they switch to an organizational focus:

...organizations need to fully embrace responsibility for correcting the current imbalance of [non-promotable tasks]. That is our primary lesson. This is not a fix-the-women problem (pp. 232-233).

Read, The No Club, then keep it close by. Follow the authors' advice to take your time before answering a request. Use that time to skim the book again. The No Club is a book for many women you know -- and for the men working alongside them.

(My tools: Google auto-complete and Grammarly, paid account)

Low-Code/No-Code: Citizen Developers and the Surprising Future of Business Applications by Phil Simon

Low-Code/No-Code is a fantastic book for all of us (this is a common theme in these summaries). We all need to do more to build our tools and craft our work. As leaders and managers, we need to make sure our organizations function by having the best talent and tools aligned with the times and toward our targets.

Contribution

Besides the engaging voice and dry humor I expect from Phil Simon’s work, Low-Code/No-Code also presents benefits for citizen developers and examples and implementation strategies for managers. Whether you take on low-code/no-code approaches as small bites or as an overarching organizational strategy, you will benefit from this book.

Aside

As these tools spread, I hope we move away from a language of “code.” I fear that thinking of this work as coding may be a barrier to some. The vocabulary train has likely left the station, but I’d rather have a language of apps.

Conclusion

Regardless of what we call these tools, if you can use a smartphone, you can be a citizen developer. Read Low-Code/No-Code to learn more.

(My tools: Google auto-complete and Grammarly, paid account)

Power and Prediction: The Disruptive Economics of Artificial Intelligence by Ajay Agrawal, Joshua Gans, and Avi Goldfarb

I enjoyed their previous book, Prediction Machines: The Simple Economics of Artificial Intelligence, but agree with the authors’ insight here -- we need more about how AI fits into our broader sociotechnical systems: the full design and outcomes of our organizations and work.

System Mindset

In Power and Prediction, Professors Ajay Agrawal, Joshua Gans, and Avi Goldfarb help us understand where AI fits into the broader picture. They show AI as a piece of an overall system and that:

"[t]he successful adoption of AI presents what we will term here the system mindset. It stands in contrast to a task mindset in that [the system mindset] sees the bigger potential of AI and recognizes that to generate real value, systems of decisions, including both machine prediction and humans, will need to be reconstituted and built" (p. 112).

Actionable Framework

They also offer an “AI Systems Discovery Canvas" (p. 241) to support leaders' thinking about AI's strategic and integrative roles in their organizations. 

My Vancouver Perspective

As happy as I am with the book's insights, I am a little sad (given I work in Vancouver) about their 2017 prediction cited in the preface: "Where do you expect to see Canada's first AI unicorn—the first AI startup to reach a billion-dollar valuation?" Our bet: "Montreal. Or maybe Toronto. Or possibly Edmonton" (p. 8). It turns out my wishes and their predictions were both off. Verafin, from St. John's, Newfoundland, hit 2.75B USD in its acquisition by Nasdaq in 2020…..

I look forward to Vancouver playing a strong role at least in the bottom-up applications of AI. My RA, Nitish Gupta, and I contribute to the basics by organizing a quarterly list of AI-enabled tools for personal automation.

Taking Action

I do my best in class and workshops to share a systems-level approach (Thinking in 5T) and will be adding the system mindset advice from Power and Prediction.

(My tools: Google auto-complete and Grammarly, paid account)

The Future of Work

Low-Code/No-Code and Power and Prediction need deeper dives than I offer here. I am not doing the significance nor the nuance justice in these short mentions. I look forward to other ways I can integrate more of these ideas into my teaching, workshops, and work. You will also find value in reading, discussing, and most importantly, applying the material.

What ChatGPT Has to Say about Power and Prediction:

I thought it only fitting that Power and Prediction be the focus of my ChatGPT test. I fed it this prompt:

Write a review of Power and Prediction: The Disruptive Economics of Artificial Intelligence by Ajay Agrawal, Joshua Gans, and Avi Goldfarb. Include page numbers for quotes. Include hyperlinks. Write in the active voice.

I have no clue what edition of the book ChatGPT is using -- the pages noted below do not match my Kindle version. Maybe it’s making them up (just like some humans I know).

In Power and Prediction: The Disruptive Economics of Artificial Intelligence, Ajay Agrawal, Joshua Gans, and Avi Goldfarb delve into the economic implications of artificial intelligence (AI). As experts in the field of AI and economics, they offer unique insights on how AI is transforming the way we do business and the way we live.

The authors introduce the concept of "prediction power," or the ability of an AI system to accurately predict the future, as a key factor in unlocking the economic value of AI (p. 3). They provide numerous examples of how companies use AI to gain a competitive edge, such as optimizing supply chains to reduce costs and improve efficiency (p. 25).

The book also discusses "creative destruction," or the disruption of existing industries and the creation of new opportunities by new technologies (p. 7). The authors examine how AI is driving creative destruction in industries such as transportation, healthcare, and finance, with a particular focus on how it is enabling doctors to make more accurate diagnoses and treatment recommendations (p. 87).

Overall, Power and Prediction offers a thorough and timely analysis of the economic impacts of AI. It is a must-read for those interested in the future of technology and its effects on society.

Links [NOTE: all of these are wrong, common problem with generative AI — use links above]:

Power and Prediction: The Disruptive Economics of Artificial Intelligence: https://www.amazon.com/Power-Prediction-Disruptive-Economics-Artificial/dp/0198805068

Ajay Agrawal: https://www.rotman.utoronto.ca/FacultyAndResearch/Faculty/FacultyBios/AjayAgrawal

Joshua Gans: https://www.rotman.utoronto.ca/FacultyAndResearch/Faculty/FacultyBios/JoshuaGans

Avi Goldfarb: https://www.rotman.utoronto.ca/FacultyAndResearch/Faculty/FacultyBios/AviGoldfarb

All in all, not a bad write up from ChatGPT, aside from the fake links. I will build and share a workflow leveraging ChatGPT for transactional writing. I also look forward to rethinking my homework assignments. More in the New Year!

Automation is part of our work. The question is whether we are finding the tools to augment our abilities or waiting for our organizations to decide what to automate and who to replace. Research finds that people who craft their work are more productive and employable. In this earlier post with Brett Li of Tonkean, we talk about that research and the steps you can take to craft your work. But what tools can help us?

Bottom-Up Automation Tools

Here is the December 2022 quarterly update of our database of personal automation tools! Nitish Gupta and I collect tools that leverage artificial intelligence (broadly defined) to support work. You likely use some of these tools already (e.g., GMail’s autocomplete). We hope you also find many valuable new tools within our mix (ChatGPT is new this quarter). Our criteria include these tools are:

• for work (again, broadly defined) - generally, not lifestyle. That said, many will help across your work and leisure.

• that individuals or teams can integrate into their work without organizational support. (Many powerful tools for automation would connect directly into an organization’s systems - these are better taken on as top-down and bottom-up efforts.

• leveraging automation either internally (e.g., machine learning or other artificial intelligence) or as part of a process connecting to AI tools (e.g., IFTTT).

Click on the image below to have a look. Mobile users may have a better experience here.

This Quarter

We added over 27 new tools this quarter. We split Photography given the size of the category and the number of new tools. Look for Image Enhancer and Image Generator. We also added a category for Chatbot. I remain surprised that all of our prior selections are still available.

What’s Next?

We update the list quarterly. Ironically, for now, this is a hand-crafted process. As we looked at earlier lists of automations for personal use, we found many were out of date. Tools may come and go, change names, are rolled into other services, and the like. We will use automation to check our links each quarter and add new tools as we find them — or as you suggest them in the comments ⇓ below ⇓.