Get clear direction with The Pilot's Compass: freely available, opinion-based research notes from Pilot Research, offered via our website and LinkedIn

Boosterism, An Historical Economic Trend With A Surprisingly Long History

I will always admit my own ignorance. As often as that occurs, for this particular note, I had thought the term “boosterism” was relatively recent. It was particularly brought into my focus by British Prime Minister, Boris Johnson; and, as some of the press labelled his approach to policy, “Boris boosterism”. Given the rabid positivity infecting the promotion of AI by the overwhelming majority of its vendors, I felt that defining it as AI boosterism was, if not new and exciting, an accurate representation of what we are witnessing. Turns out boosterism has been around as a term since the expansion west of the US population, with cities and regions engaged in overstating their advantages in order to attract people.

This is true, to some degree, of all emerging technologies; however, in the case of AI the scale of what’s at stake is much, much larger. I wrote about AI being “... too big to fail,” in another Pilot’s Compass note. If on reading that if you believe that even some part of my argument is true - that it’s not just money, but effort and reputations too - then we’re witnessing boosterism on a level not yet achieved by technology (okay, maybe nuclear energy, I mean, according to the mid 20th century, we were all meant to be driving around in nuclear-powered cars… right?).

Defining AI Boosterism? Ask Who Benefits

Overselling the, perhaps imagined, infrastructure of cities to attract settlers is a lot like the boosterism that pervades AI currently. How can we define it?

• Unbound optimism - there can only be radical, even seismic, positive change to the way we do business, live our lives and, all of this is good: without question. Worried about jobs for people? No problem! New jobs we can’t currently think of will emerge to create work for people. That approach always works out, just ask artisan manufacturers after mechanisation in the industrial revolution.

• Risk? What risk? - the risk of adopting the latest, greatest technology is, well, non-existent because, if you don’t, you’re a dinosaur in the path of a meteor shaped by all your competitors that do adopt.

• Promotion without question - act now! Don’t miss this opportunity! You’ve been selected (okay, that last one may just be my spam email folder). Creating urgency around purchase, implementation, and use only echoes the old saying, “Act in haste, repent at leisure.”

• Profit first - let’s be honest with ourselves, who is pushing the sale and use of a technology? The answer is people who are motivated to sell it. Does it work? Is it appropriate? Does it align with our values, standards, mission or, the sales quota being chased by the rep?

• Solutions will solve everything - don’t worry about it, technology, or, something else, will resolve all those issues that the technology you're using may / will create. An unhappy workforce? We’ll find new work for them, or, they will be provided for by some new opportunity we’re, ahem, currently unable to define.

All of this sums to an almost incalculable (perhaps we should ask AI?) level of fear of missing out (FOMO) that is forged, stoked, and promoted by those who fall into one (or both) of two camps:

1. People who have invested money, effort, and reputation into the technology - those people who have engaged with what can / should be identified as AI boosterism. There is too much on the line, failure is not an option in their narratives.

2. People who stand to gain, personally, from “selling” the technology in question - the obvious example here is sales professionals, but it’s not just sales that stand to gain. Those who market, opine, and provide advice on the technology are all potential boosters.

Bodybuilding May Have Become Relevant To Technology: Cuts Versus Gains (So I’m Told)

I cannot and will not deny the potential gains of AI-powered automation. They are of the exact type as any other technology we use to improve productivity; however, we must also accept there may be a human cost associated with that productivity gain. Currently, I assert that the majority of AI use cases that actually deliver measurable return are the result of one of two factors:

Cuts: AI technologies displacing human effort therefore reducing cost - the next iteration of robotic process automation (RPA), AI and AI agents take on work that traditionally was paid human labour. Just like almost every technology that has had a measurable economic impact, from agricultural machinery to manufacturing automation.

Gains: highly specialised high-data, high-speed use cases that humans cannot achieve - from high-speed trading profiting from tiny differences in markets to mass personalisation of online marketing, there are clearly some things which humans can’t do.

The potential applications of AI are phenomenal, for example examining medical scans to assess anomalies at speed and scale, to finding new blockbuster drugs. But there is little balance in the conversation about AI and if you entertain the astronomical highs of potential good: you must equally entertain the opposite, negative outcomes.

Like Agent Mulder, I Want To Believe

I want to believe that AI technology will improve outcomes for everyone, but in its current state, the scope of its benefits is pretty much the same as any other technology. AI technologies either serve cost-cutting use cases which will squeeze human labour or meet use cases that only exist as a result of previous technology adoption. In both of these cases, it will make some people who are incentivised to promote and sell it wealthier, and in cost-cutting uses take away work from many other people who have no obvious path to sustaining themselves and their families.

As a buyer or user of AI technology, perhaps instead of falling head-first into FOMO, we should work to understand how it could improve outcomes for you, your organisation, and all those people who work with - and buy from - you. Technology for technology’s sake is not a good answer to any serious question (and it pains me to say that, I love technology).

By first identifying the need and how it benefits your people, be them employees, partners, or customers, whether it’s AI or a new coffee machine, technology acquisitions will pay back bigger, demonstrable returns.

Whatever Direction Your Compass Points…

Thank you for reading this Pilot’s Compass note. It will be available on both Pilot Research's website and LinkedIn. Is it possible to tackle this subject in around a thousand words? Of course not. My hope is that it’s brief enough to take the time to read and - agree or disagree - it perhaps proved somewhat thought-provoking. I welcome your comments, feedback, and ideas at tom@pilotresearch.co.uk

Get clear direction with The Pilot's Compass: freely available, opinion-based research notes from Pilot Research, offered via our website and LinkedIn

A Simile Too Far, AI Is Like Ducks: Masters Of Calm On The Surface Supported By Furious Effort Beneath

I live in an area of the United Kingdom famous for, amongst other things, its definitional Aylesbury duck. The Aylesbury duck is found on the town’s coat of arms, locals are known to call themselves ducklings, and (approve or not) you’ll find it on the menus of high-end restaurants around the world. In what might be the most ambitious of my technology-focused similes to date, AI technologies are - to the user - like the gliding magnificence of an Aylesbury duck swimming across the surface of a twinkling village pond. Supporting this graceful simplicity, beneath the waterline, are large, webbed feet moving at pace, propelling it through a murky, low-visibility environment where weeds and debris obstruct its otherwise graceful progress.

Hopefully, at this point you might have an idea where this line of reasoning is going. Either that, or you think I’ve entirely lost my grip on reality. Humans are “designed” to interact with other humans, not tabs, columns, cells, keyboards, mice, or… and so on. The rise of Large Language Model (LLM) -powered AI assistants and yes, fine, AI agents, is the triumph of human-native interaction versus the long legacy of computing and its various forms of interaction, input, and commands. But it’s more than just the user interface (UI).

History Repeats Itself Creating An Ever Bigger Problem

For those with longer memories, concepts in business IT such as the Enterprise Service Bus (ESB), Service-Oriented Architecture (SOA), and more recently, the ubiquity of Application Programming Interfaces (APIs) know about the challenge of tackling ITs legacy. That legacy comes in many forms, from outdated systems, critical financial and resource management systems and banking platforms that were originally conceived decades ago, or a host of cloud-based software-as-a-service (SaaS) solutions that should - but don’t always - talk to each other. For good measure, add to this the “big data” created by all of the above and the Internet of Things, alongside a generous helping of the Internet and you have the ultimate construct of disparate data generating systems in a massive, growing, interdependent, and heterogenous digital monstrosity.

Humans created the technology, the technology does what humans can do but far faster, at a greater scale, and not aligned to some golden international standard: we have a technology-created problem that only technology, in this case AI, can overcome.

“Early Grey, hot.” It’s Called Natural Language For A Reason

Returning to ducks, let’s say above the water is the AI interface, plain text or speech (natural language for those with an IT hat to wear) which is easily understood, approachable, and does what it’s expected to do. Beneath the waterline, lies the murky depths of data and systems to be navigated to produce the effortless appearance above.

At this point the simile fails, spectacularly, given that pond water is crystal clear by comparison to the ever-changing spaghetti junction of software and data that is the fog of war enshrouding modern technology. But the idea, at least in my view, stands - AI in the form of LLM-powered AI assistants / agents / interfaces are the first part of the final step in the evolution of how we interact with technology and data.

Why? Because in 99.9%+ of cases, most people do not need or want to know how the web of software, data, and computing hardware that sits behind the application they’re using works. Or, put another way, behind the interface it could be an army of guinea pigs running a treadmill-powered mechanical doohickey, as long as it comes up with the solution.

Simplicity Hides Dangers Lurking Beneath The Surface

If you’ve listened to me speak or read my work you’ll know I regularly say, “What could possibly go wrong?” The answer to this question in this case, is: quite a lot. With the rapid adoption of a technology that hides the mass of data it's trained on and accesses to augment its answers, while also acting as digital filler to connect disparate applications and services, a number of immediate concerns arise.

I suggest that primary amongst these is that we are running the risk of creating a solution of systems so complex and obscured that no one can understand how it arrives at the answers it does, how it works, and therefore how to effectively maintain it. AI may well be the new UI that helps us navigate myriad systems and data but how much confidence and trust do we have in its outputs? How do we check its work? When an underlying system is updated and changes its process or data output, perhaps fails or is compromised, or the data itself is corrupt / wrong / insufficient, how do we know with certainty? Our ability to create technology looks to be surpassing our ability to effectively manage it.

Of course, this raises other questions about the degree to which we rely on the output of AI and take responsibility for action taken upon it by either humans or machines.

Ducks, Dinosaurs And The Importance Of Explicability

Currently I’m working on a framework for evaluating trust in AI solutions and I keep coming back to a core principle. That is, can you explain the inputs, working, and output of the system in a way that anyone, not just someone with a PhD in computing science, could understand?

I don’t know how closely related ducks are to dinosaurs, but to borrow from the film Jurassic Park I don’t care whether the system has two million lines of code, or a trillion plus. Even if it does what it should do, effectively and efficiently (measured against the value of what it is producing) then it’s already scaled well beyond most people’s ability to understand it. Like a duck, or dinosaur for that matter, though, can I explain it without necessarily understanding the minutiae of each individual process and mechanism that makes it work?

If yes, there’s hope. If not, I strongly suggest we’re into territory that starts to align with another favourite saying of mine: just because we can, should we? We don’t all have advanced degrees and doctorates in computer science, mathematics, and data science. But, somehow, we are already and increasingly subject to the output of the systems these disciplines create.

Bottom line? I’m forced to borrow from the quote, “If you can’t explain it simply, you don’t understand it well enough.” Whether that was Einstein or not, the question that hangs over the apparent simplicity of AI as an interface to our complex, growing collective of systems and data is clear:

“If you can’t explain it, should you really depend on its output?”

Whatever Direction Your Compass Points…

Thank you for reading this Pilot’s Compass note. It will be available on both Pilot Research's website and LinkedIn. Is it possible to tackle this subject in around a thousand words? Of course not. My hope is that it’s brief enough to take the time to read and - agree or disagree - it perhaps proved somewhat thought-provoking. I welcome your comments, feedback, and ideas at tom@pilotresearch.co.uk

Get clear direction with The Pilot's Compass: freely available, opinion-based research notes from Pilot Research, offered via our website and LinkedIn

Digital Labour: The Latest Solution To All Problems

Digital Labour is, I’d say, impossible to avoid as a topic in the technology world. This is the latest iteration of a broad narrative, that the ever-present challenge of growing / maintaining / stopping a collapse in (delete as you see fit) productivity, can be solved through the use of technology. Very particularly, in this case of this conversation, digital labour is being brought to us in the form of AI “agents”; leading to the - in my view - awful term, “agentic AI’.

I believe agentic AI (sigh) and the push to digital labour is simply the latest iteration of the well-established “more technology equals more productivity” narrative that has been around as long as, well, technology has. But, not just technology in its digital form; from wheels to water mills and steam engines to quantum computers, technology continuously changes the way we live and work. In terms of how we define an AI agent, there are competing definitions, but for me the simple way to think about them is as a natural progression of AI assistants. AI agents are:

• Given goals to achieve, usually within more complex or multi-step processes;

• Empowered with greater ability to take action based on available information;

• Require no or low human input; and,

• Likely have learning built-in to improve performance.

In other words, more human-like, hence, digital labour. So far, so good.

History Teaches Us That Technology Adoption Often Leaves Humans As An Afterthought

The economic historian in me continues to tug at this train of thought, however. Let’s keep things fairly macro. The industrial revolution displaced huge swathes of the workforce, but did create many new factory jobs in cities and the supporting industries, for example, coal mining. The shift from industrial to services-based economies continues in many regions, but also disrupts workforces while creating social upheaval. In this case, let's take the coal example again and look at what has happened to many communities built around the mineral extraction industry. In each of these cases, the technology results in transformation, but not for the benefit of society-at-large, but rather technology’s sake (and the interests of those who build or buy it).

I promise I have not travelled in time either literally, or via literature, and been converted by Mssrs. Marx and Engels. Just, in each of the major shifts powered by technology, I assert that the human aspect has been, well, an afterthought.

Humans Aren’t Just Your Employees, They’re Your Customers, Partners, And Investors Too!

At risk of being worn out, various combinations of “People”, “Process”, and "Technology" have been staples of the practical guide to technology adoption. It is well-worn for a reason though, because addressing only one or two of these things and ignoring the remainder is, often, a major contributing factor to failure. or we could be generous and say limited success when it comes to delivering technology projects and programmes.

My suggested success formula takes these into account, and adds a little more:

Productivity Investment Return = (Software Tooling + Process Optimisation) X (People Upskilling + Employee Engagement + Corporate Culture)

I contend that the software tooling (technology) and, likely, the process optimisation is assumed by most; whereas the multiplier - investment in people, engagement, and culture - is quite possibly not.

• Scenario 1: Invest in software tooling, but not process optimisation - make inefficient, or worse, wrong stuff happen really, really fast;

• Scenario 2: Invest in software tooling and process optimisation, but forget about training people - frustrate the workforce with challenging learning curves that damage productivity. And / or people will find fun “workarounds” that break processes (law of unintended consequences, anyone?); and,

• Scenario 3: Invest in software tooling, process optimisation, training, but… disregard engagement and culture - watch as your investment drives employee disengagement, burnout, and quiet quitting (amongst others…).

While each of these scenarios could be considered extreme, and perhaps are, none of them look particularly good outcomes for people. Those people might be your workforce, they could also be your customers or partners. Would you wish to impose any of the above on them? If so, that’s a conscious choice and although it wouldn’t be one that I’d make, I get it.

Digital Labour Cannot Fill The Gap Created By Underinvesting In Humans

When it comes to AI agents, as an employee I believe it is entirely justifiable to worry about the impact this latest, greatest technology is going to have on my prospects. When machines can start doing human work, humans have to find something else to do. I’ve written about this before (2019 was my last piece on the subject) and when I did, I said,

“... the problem of technological unemployment bristles with new AI-powered possibilities for putting people out of work, while radically concentrating wealth in the hands of a very small number limiting the chances of new jobs/job types, services, and products rebalancing the economy.”

I still believe this conversation does not attract the serious attention it clearly deserves. I also believe this is not an either / or conversation, but to realise that means investing in people too.

Whatever Direction Your Compass Points…

Thank you for reading this Pilot’s Compass note. It will be available on both Pilot Research's website and LinkedIn. Is it possible to tackle this subject in under a thousand words? Of course not. My hope is that it’s brief enough to take the time to read and - agree or disagree - it perhaps proved somewhat thought-provoking. I welcome your comments, feedback, and ideas at tom@pilotresearch.co.uk

Get clear direction with The Pilot's Compass: freely available, opinion-based research notes from Pilot Research, offered via our website and LinkedIn

AI Investment Is Not Just Money And Is Multiplied By Fear Of Failure

Recently I said on a podcast, “I’m getting ‘too big to fail’ vibes around AI’.” The two obvious thoughts may be:

1. Tom Pringle talking about “vibes” makes me wonder if it’s actually a deepfake; and,

2. Really? It’s not like AI is the same as a bank, the kind of organisation which the phrase is most often associated with, if it doesn't work well… it’s not like the economy stops working… right?

On the first point, I can assure - but not prove - it isn’t. On the second, I think there is simply too much invested in AI to allow it not to become “successful”. Let’s take a look at what’s invested in AI:

• Money - like Dr. Evil, I shall raise a pinky finger and say, “... one billion dollars!” Except, you need to add many hundreds of billions of dollars, euros, pounds, and so on. The cheques (yes, I’m British) written against investment in AI startups, infrastructure, product development, and organisational projects sum to unimaginably huge numbers. A trillion dollars? Sure, I’ll see your trillion and (entirely guess) it’s more if you include planned and announced spend.

• Reputations - reputational capital is important, for some it may even be at least as important as financial capital is. A lot of senior software executives have taken to the digital megaphone to announce that the age of AI is upon us and its impact will be fundamental on both our economies and societies. It’s not just tech execs and investors though, politicians, consultants, analysts, and that guy you see at the pub sometimes are all (fairly) certain that AI is going to remould the foundations of our economic and social fabric.

• Effort - take the financial and reputational investments and add the volume of human effort and power already ploughed into AI and you have another significant sunk cost. People, myself included, are learning by using and taking courses to understand how AI tools may be applied to work. We want payback for this effort, be it for personal use or corporate advancement.

The opinion I’m offering on AI investment is that it is more than simply money. Put otherwise: Investment in AI = Money + Reputations + Effort. I suggest that calculation is taken and multiplied by a very difficult (impossible?) to quantify variable: fear of failure.

Hopefully, at this point in our conversation you will at least, if not agree with, what I’m suggesting: AI technologies have to be seen as successful, even if that means we have to think about "success" a little differently. The stakes are simply too high.

Step Back To Realise Value And What It Means For You

Hype? Bubble? Demonstrable Value? All, and more. Why choose when as a friend of mine once said, “Two things can be true at the same time.” In this case, I think it’s more than two things, there are obviously multiple sources of value in AI technologies. I don’t have to rely on studies and data points to support this much beyond my own, immediate experience. While limited, that experience has proven that the everyday of my “knowledge work” at least AI can be a very useful time saver when it comes to collating research, or creating an image to support a website post.

What of AI’s potential beyond the day-to-day of productivity, such as healthcare advancements and scientific breakthroughs? Undoubtedly possibilities that go beyond the modest scope of this note, but here I’m focussed on everyday business value, not re-writing the social and economic fabric of our world. Does it mean I’m missing the point of AI? I will freely admit, maybe yes, but given the oxygen-out-of-room-sucking effects of AI on investments of money, reputation, and effort in alternatives, the pressure to experience short term benefits is close to explosion point in my view.

Therefore the question that hangs in the air for me, at least, in its current state will it repay its total investment? Currently, I have to say, “probably not.”

This is not an anti-AI tirade. I could offer such a rant based on my love of science fiction, its often persuasive predictions, and my fondness for what I like to call techno-philosopy. But forget all of that, I believe two things need to happen for AI to be successful in the near term:

1. We must turn down the heat - some of the white heat of expectation that pervades AI needs to be removed. The greater the amplification, the greater the expectation. While I don’t doubt it’s possible for AI technologies to deliver this in the long(er) term; although we may not be dead (sorry, J.M. Keynes), the fog of uncertainty grows ever-thicker the further we look out.

2. Understand what meaningful value from AI means to you - let’s focus on the things we can see at least somewhat ahead. Value means a lot of different things to different people, of course. Investigating what it means in the context you find yourself in, be it corporate, charity, governmental, or personal is key. What can you see and achieve (nearly) here and now with AI?

Whatever Direction Your Compass Points…

Thank you for reading this Pilot’s Compass note. It will be available on both Pilot Research's website and LinkedIn. Is it possible to tackle this subject in under a thousand words? Of course not. My hope is that it’s brief enough to take the time to read and - agree or disagree - it perhaps proved somewhat thought-provoking. I welcome your comments, feedback, and ideas at tom@pilotresearch.co.uk 

The role of data in AI should never be underestimated. Where data is the input, the output will be subject to how good, or not, that data is. In previous papers I have written for the Artificial Intelligence Trust Quotient (AT-TQ) I explored some of that role: primarily as the raw material used for training AI models. As I have tried to argue, while the technology and techniques behind the scenes may be extremely complicated, the principles of what is happening are surprisingly easy to explain.

Simplifying complexity is also the approach of the AI-TQ framework. By examining the nature of inputs, the approach taken to processing them, and the nature and impact of outputs, assessment of AI solutions is possible without delving into the nuts and bolts of the technology.

The Power Of Unintented Bias

The “traditional” approach to programming a computer to do tasks is rules-based. For example, if this happens, do that and if that happens, change this, and so on. In AI, we cannot create a rules-based system for, say, recognizing what the subject of a picture is, for example, a dog. In the case of machine learning, we give the machine example pictures of a dog and the model works out the probability of what the subject of new images is; the greater the volume of (accurate) training data the better the model works.

But imagine this, perhaps my job is to prepare the data for training a model to identify dogs in pictures. On a personal level (in other words, bias) I have always had a fondness for sighthounds such as Greyhounds. While preparing my training data (pictures of dogs) for the model I, quite possibly unconsciously, favor sighthounds over pictures of other breeds of dogs. The AI model is only as good as the data used to train it, so in my example above the model may turn out to be extremely good at identifying Greyhounds, but pretty rubbish at identifying, say, Pugs. Here is the problem, AI can only make decisions based on what it has been trained on and what it is trained on is subject to any (intentional or otherwise) bias within the training data.

Bias (in its varied, subjective forms) in data sets is highly probable for many reasons, not least given changes in societal norms we experience over time. It may also be the case that the bias in a dataset is unacceptable legally, societally, or for a particular organization. In the case of data sourced from the internet as is often used with Large Language Models (LLMs) we can find many different - often diametrically opposed - points of view, each of many perspectives asserting they are “the truth”.

These ideas in mind, the question becomes simple:

“If the data used to train an AI model contains information we don’t want to influence the AI’s outcomes, for example, sexuality or ethnic background either by design or accident, how do we manage this?”

When Mostly Harmless Could Become Very Harmful

Data or information that represents identifiable elements of who we are should be treated with the highest respect. The definition of Personally Identifiable Information (PII) is well-established* and broadly applied through regulatory requirements such as the European Union’s General Data Protection Regulation (GDPR).

Back in the bad old days of big data** some observers liked to talk about people creating a “digital exhaust”, I prefer to think of a “digital wake”. As we move through the digital universe, our activity creates waves of data that expose elements of who we are, what we do, what we like and dislike, and so on. Whether exhaust or wake, this data can be used to build increasingly accurate and sophisticated models that represent us. The primary driver to their development was to drive digital advertising revenue and while potentially annoying to the subject, given the limited scope of outcome, serving badly targeted ads, I believe people would consider them mostly harmless.

Extend the scope of those outcomes with AI making wider-ranging and more impactful decisions and that mostly harmless data could drive real harm. Harm that could be founded on some of our most fundamental personal characteristics.

Identify The Inputs To Manage Risk In Outputs

The AI-TQ does not attempt to decide or dictate to users what personal characteristics mean to them. It does offer a range of widely-recognised measures to consider as part of its customizable approach. In this case, the variables are simply the presence of a type of data / content used as input to the AI model training or source data for AI models in use. Users of the AI-TQ will weigh the importance of different variables according to their views and requirements. In my opinion, generally the more of these data types present the higher the risk.

Some basics of personal information are provided as a starting point. These include widely recognized content types such as names, unique identifiers and location data. Biometric data types are included as part of the assessment’s offering.

The AI-TQ also includes a set of variables I call “possible discrimination drivers”. These are usually considered some of the most private of personal information and could include things like, medical history, ethnicity, disabilities, sexual orientation, medical history / records, and religious identity. Without doubt, and as with any section of the AI-TQ, the organization / individual using the assessment tool, what constitutes these data types and their relative weight is open to a great deal of interpretation.

• Personally Identifiable Information

• Unique Identifiers, e.g. Passport Number, Driving License Number

• Name

• Physical Address

• Location Data

• Country

• Region

• Precise

• Online ID, e.g. IP or Cookie Data

• Biometrics

• Any, e.g. Facial, Eye / Iris, Fingerprint Scans, DNA profiles

• Potential Discrimination Drivers

• Gender and Gender Identity

• Ethnicity

• Religion

• Medical History / Records

• Political Affiliation

• Education Level and / or Attainment

• Sexuality

Coming Next To The AI-TQ: Regulation

This research is part of a series that will culminate in the official launch of the Artificial Intelligence Trust Quotient (AI-TQ) assessment. The final part of this series comes next, covering the subject of regulation of AI, its sources, uses, and monitoring.

In addition to my gratitude for those providing feedback and support throughout this process, I’m excited to launch the framework on a freely available basis in the coming weeks.

* See the UK’s Information Commissioner’s Office definition here https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/personal-information-what-is-it/what-is-personal-information-a-guide/

** Data has always been “big” in respect of our ability to collect, store and manage it

To state that the purpose and intended outcome of an AI-powered solution is critical to building trust in it, seems obvious. That’s because it is. But quantifying the nature, scale and impact of an AI powered solution is - to say the least - tricky. Recent advances in legislation help highlight this point, with the new EU AI guidelines amongst the first in the world to attempt to bring this into a more detailed, regulatory framework.

The purpose of the Artificial Intelligence Trust Quotient (AI-TQ) is to help non-technical business stakeholders understand and therefore decide their level of trust in AI-powered solutions means it must consider the imperative driving the use or development of those solutions. Introducing the assessment areas of Purpose and Outcome, the AI-TQ is trying to help its audience navigate what is potentially a complex topic; and, one in this case that may be shrouded in an additional technical language, law.

Purpose + Outcome = Impact

I believe it’s fair to say that both these areas of assessment, Purpose and Outcome, are so significant they warrant their own, dedicated coverage. So why combine them?

Together, Purpose and Outcome are more than the sum of their parts, I think of it as being similar to a simple equation. For example, if the purpose is commercial, say tailored advertising, and the outcome’s scope is limited to offering a small discount for a product or service then the result of that equation, the impact, is likely quite limited. At the other end of the scale, perhaps the purpose is making decisions about an individual’s entitlement to a form of life-extending healthcare which would have an impact immeasurably greater.

Serving A Commercial Or Civil Purpose

As with most forms of assessment, it is possible to substantially overcomplicate what defines the purpose of an AI-powered solution. To keep things simple, the AI-TQ splits purpose into two broad areas:

• Commercial - commercial purposes will be focused on the generation of revenue, management of costs, or to meet compliance and risk management requirements in the private sector.

• Civil - civil purposes will be focused on improving the services and outcomes delivered and sought by public sector (state, federal, local government and related organizations) bodies.

Without judgment, it is important to note that AI solutions used for purposes that include military, battlefield and other defense activities funded by either / or private or public organizations are specifically out-of-scope for the AI-TQ.

Outcome Requires Additional Perspectives

Different outcomes and decisions have different magnitudes of impact, impact on different groups or categories, and may, or may not be automated. The AI-TQ seeks to assess the type of outcome in terms of the Purpose variable, whether it is commercially or civilly focused, and impacts an individual, a group, or an organization.

The framework also assesses the nature of the outcome. This is relatively simplistic in a determination as to whether the outcome is reversible or not.

Automation is also a factor in the outcome. If the outcome is automated, with no obvious route to review or monitoring for erroneous output covered by the “Human Governance” assessment area of the AI-TQ.

• Individual

  • Low Scale of Impact, e.g. automated personalization of web content.

  • Medium Scale of Impact, e.g. automated tailoring of service offering.

  • Highest Scale of Impact, e.g. automated decision making with high degree of impact on subject’s outcomes.

• Groups

  • Low Scale of Impact, e.g. automated personalization of web content.

  • Medium Scale of Impact, e.g. automated tailoring of service offering.

  • Highest Scale of Impact, e.g. automated decision making with high degree of impact on subject’s outcomes.

• Organization

  • Low Scale of Impact, e.g. automated personalization of web content.

  • Medium Scale of Impact, e.g. automated tailoring of service offering.

  • Highest Scale of Impact, e.g. automated decision making with high degree of impact on subject’s outcomes.

How users of the AI-TQ choose to define low, medium, and highest impacts will, of course, be open to interpretation. In combination with other assessment areas, such as “Values” and “Human Governance” it is the aim of the AI-TQ to provide an overall assessment that combines different perspectives on the technology in question.

Coming Next To The AI-TQ: Personal Characteristics

This research is part of a series that will culminate in the official launch of the Artificial Intelligence Trust Quotient (AI-TQ) assessment. Next in the series is “Personal Characteristics” - perhaps one of the most hotly-debated subjects within the AI arena, an examination of the use of data in AI-powered solutions that helps identify us.

Also, my continued thanks and appreciation for the feedback and comments on this research so far! It is immensely valuable to me, and I look forward to more.

Industry Analysts And Independence

I know I’m not alone in my belief that being an analyst in the technology world means being rigorously independent. But what does being independent mean?

Of course, there are many different interpretations and definitions of what it means to be independent, and we could venture down a fun, philosophical conversation. I’m not sure how valuable that would be though. My interest is in a practical, shareable and disclosable perspective that gives a starting point from which people can then make up their own minds.

A Simple Starting Point

Independence in my view is having no financial incentive that could influence my work that is undisclosed. Beyond that definition it is up to others to decide whether what I’ve written, presented or talked about is independent or not.

I’m making four commitments to back up my view:

1. Pilot Research will maintain an “Independence” page on its website which discloses any technology company shares / stocks, options or related financial instruments held or directly controlled by either the company or its employees.

2. All content produced under contract and published directly by Pilot Research will be explicitly marked to indicate it is paid for.

3. For content produced discussing event attendance (vendor or otherwise) where travel and other expenses have been reimbursed by a technology company will be specifically highlighted in that content.

4. Finally, all paid for software (including software received for free that would usually be paid for) used by Pilot Research and its employees for the purposes of conducting its business will be listed.

I am sure this list will evolve and I welcome any comments or suggestions you might have about how it could be improved.

The next time you’re reading or listening to analyst content it might be worth thinking about how they define independence and disclose information.

No matter how well established, proven, and tested a technology may be, there is a - often unspoken - trust dynamic that governs our acceptance and use of it. For example, selecting the “Auto” setting on a dishwashing machine and trusting it to do a reasonable job of deciding how to wash those items seems perfectly okay to most. At the opposite end of the scale, thousands of aircraft daily are safely flown by computerized autopilot to their destination; but as a passenger, although you likely trust the technology onboard to do its job, would you be happy to take that flight if there was no human pilot in command? Of course, some will answer “yes” to this question: but I suspect many more would not.

The examples above highlight two key factors in how we typically gauge our acceptance of a technology; first, the nature of the outcome; and, second, the presence of human governance. Given the nature and scale of the outcome is such a formidable driver to trust, it has its own assessment area in the Artificial Intelligence Trust Quotient (AI-TQ). Human governance or oversight is, in my view, often overlooked yet critical to building and sustaining trust in the development and use of AI.

Human Governance Of AI Should Not Be Taken For Granted

Including human governance in the AI-TQ is to purposefully consider the inclusion, role, and influence of people in the decision process of an AI powered solution. Inclusion is guaranteed to some degree with all AI solutions given at some point people have been involved in its development. As the software is deployed the inclusion of people becomes less certain and may tend toward technical management; that is the on-going maintenance of the solution at a technical (are the lights on?), rather than holistic (are the lights on in the right place and at the correct brightness) level.

Is The Person Just Informed? Or Empowered To Act?

Human-in-the-loop is a well-worn phrase amongst many in the AI technology world, but what does it actually mean? The answer is, as with so many technology definitions, “it depends.” While its implication is clear, similar to the autopilot example above, what it means on a case-by-case basis and to what extent that human is simply informed as opposed to empowered to influence or act is critical.

The AI-TQ attempts to specifically define what it means to have humans in the loop, as creators, users and subjects of the technology and its use. This last group, subjects of the technology’s outputs, are paramount to building trust in the technology.

Trial By AI Jury

While current AI solutions have a relatively limited scope of outcome, there can be little - in my view no - doubt that the scope and scale of the outcomes powered by AI will grow exponentially. Let us consider another analogy to highlight the importance of human governance. Typically, democratic countries provide their citizens with the right of trial by a jury of their peers. I don’t intend to delve into the political philosophy behind this approach, but rather use it as a way to highlight that a potentially highly technical topic, in this case the law, may not always produce a result considered acceptable by those subject to it. Of course, legal professionals are involved, but the decision made in the end is in the hands of people, rather than an uncompromising interpretation of a technical text that cannot encompass all possible scenarios.

This is, possibly, a more palatable exploration of the idea than the usual “thick end of the wedge” from which we work backwards to the current day. Typically that future involves machines deciding that what’s best for humanity is not to let humanity make the decisions. Extreme, certainly, but as my Mum is so often fond of saying, “never say never.” Science fiction fun aside, it may be the case that a small, self-appointed group emerges to control the technology for their benefit, rather than the majority’s. A worrying possibility with more than a hint of plausibility: making the importance of broader, human governance essential.

Human Governance In Human Language

Three different elements explore the inclusion of human governance of AI-powered solutions in the AI-TQ:

• Right Of Review - does the organization who operate the AI solution offer those subject to its outcomes the right to request an explanation of how that outcome was derived and the ability to have a person review it?

• Written Commitments - does the organization who operate the AI solution publicly offer written terms, conditions and / or written commitments that specifically cover the operation and use of AI solutions? Are these written in a way that a non-expert (technical or legal) can understand?

• On-going Oversight - is there a formalized governance body at the organization that operates the solution which includes both technical, business, and customer stakeholders?

Coming Next To The AI-TQ: Purpose And Outcome

This research is part of a series that will culminate in the official launch of the Artificial Intelligence Trust Quotient (AI-TQ) assessment. Next in the series is “Purpose” and “Outcome” - combining two critical assessment areas that investigate the scope and scale of an AI-powered solution’s potential impact.

Also, my continued thanks and appreciation for the feedback and comments on this research so far! It is immensely valuable to me, and I look forward to more.

Explaining AI: Magician Not Required

Technology can be, well, pretty technical. Even in what appears a simple use case to the user, what is going with software, hardware and supporting technologies may well be darn complicated. Artificial intelligence (AI) technologies are an excellent example of this issue, with words like “magic” often used to describe the experience

As a fan of science fiction, Arthur C. Clarke’s third law, “Any sufficiently advanced technology is indistinguishable from magic.” Seems an appropriate quote to underscore the point. It is, however, very unlikely that magic will work well as the explanation when it comes to justifying the development and use of AI.

Explicability Is Desirable And Demanded

Being able to explain how an AI-powered (or any other technology solution for that matter) came to a decision would appear to be a reasonable request. If, for example, an insurance company was using AI to make decisions about whether to offer cover to people, it should be able to explain to a customer why cover was offered or refused. Not being able to explain it in a way understandable to a customer who is likely not an expert in AI creates obvious risks. Poor customer service is clearly bad news, and with existing and emerging regulation that governs data, analysis and AI use, bad news could become catastrophic.

The focus of the Artificial Intelligence Trust Quotient (AI-TQ) is the business users of AI powered technology. Their interest in being able to explain to the subjects of AI-powered decisions is clear, but there is an obvious gap that lies between the technical explanation of AI, open only to experts, and the regulatory explanation required by compliance experts: do I trust it?

Black Boxes Do Not Belong In Business Technology

When trying to explain how something works, transparency is - perhaps obviously - important. The question that needs to be addressed is,

“If an AI solution being developed or adopted is a black box, whether by design or perhaps shielded by complexity, then how is it possible to trust it?”

For the purposes of this research, consider a technology “black box” simply as the inability to see how a solution works. Or, in other words, its inputs, processes and outputs are not transparent to the point of being open to explaining how they work together.

Technical Complexity Does Not Preclude Transparency

Just because technology can be complex does not mean it is impossible to explain how it works. For example, while I don’t know all the technical details about how the engine in my (admittedly nearly 30 year old) car works, I can definitely understand the principles, inputs and outputs which make it work. Extending this type of understanding to far more complex systems is entirely achievable and one of the driving forces behind creating the AI-TQ.

If popular scientists are able to explain some of the mechanics of the universe in a way that we can all grasp, surely we don’t all need doctorates to gain an understanding of the fundamental principles that power an AI solution?

Defining Something That Isn’t There: Transparency

The AI-TQ offers six areas of assessment to investigate the degree of transparency in an AI-powered solution. They are designed to explore the extent to which it is possible to see and understand how the solution works from a non-technical perspective.

• Auditability - is it possible for external parties to examine the process, input, and output steps taken by the solution in coming to its outcome?

• Explainability - can the solution and its decision logic be described in a way that a non-expert would understand?

• User Experience - is it clear to someone using the solution that they have been subject to an AI-powered solution? For example, explicitly labeling a AI-powered chatbot as such, as opposed to suggestion, including by omission, that it could be a human answering the user’s questions.

• Documented Development Process - is there a clearly documented development and maintenance process which covers:

  • Business / product requirements

  • Technical development process

  • On-going maintenance and support of the solution post implementation

• Ecosystem and Community - is there an open and active community that engages with the adoption / use of the solution? Is there an ecosystem of partners who work with the solution?

• Open Source - is the solution and its components on an open source license that enables third party interrogation of the technology?

Coming Next To The AI-TQ: Human Governance

This research is part of a series that will culminate in the official launch of the Artificial Intelligence Trust Quotient (AI-TQ) assessment. Next in the series is “Human Governance” - an exploration of the importance of accessible, human oversight of AI solutions.

Also, my continued thanks and appreciation for the feedback and comments on this research so far! It is immensely valuable to me, and I look forward to more.

Data: The Fuel Of AI That It Would Be Foolish To Ignore

Many of those in the world of data, business intelligence and analytics are very familiar with the line, “garbage in, garbage out.” This is just as true of artificial intelligence (AI) technologies. In fact, given the broad scope of use cases and outcomes that could have AI applied to them, output may surpass garbage status and become toxic waste.

It’s my - sometimes unpopular - view that, largely, AI is closely aligned to or perhaps even simply a part of the (on-going) evolution of data and analysis technologies. From the early days of decision support and management information systems, on to business intelligence, self-service and predictive analytics, and, prior to Generative AI (GenAI), data science and machine learning (ML, GenAI’s less trendy sibling). It has long been known that a lack of data, poor quality data, or unrepresentative data fed into analysis will at best lead to low quality results, at worse, bad decisions that run the risk of harm.

AI Is Data-Driven At Speed And Scale

That AI can make decisions at speed and scale, and potentially that are not reviewed by humans makes the risks of bad data a lot worse. The risks to brand and reputation are obvious, an easy example being an AI-powered chatbot on a consumer website giving inaccurate, misleading, or even offensive answers. The risks of emerging use cases, where AI is tasked with making decisions that impact outcomes for individuals, groups and organizations (the AI-TQ covers this as its own assessment area, Outcomes), are substantially greater.

Trust In Data Is More Than Technical Accuracy

The quality of a data-driven decision depends on the quality of the data used and leads us quickly to questions of trust. Fundamentally, do you trust the data? When we use the word trust do we mean that it is technically accurate, or that we are confident it is both accurate and sourced responsibly? This question is at the heart of the AI-TQ, just because something is technically correct, doesn’t automatically mean we necessarily trust it. Perhaps the source data is from a third party with questionable data practices, or gathered without knowledge or consent of participants?

AI Is Driving The Agenda But It Won’t Get Far Without Data

Over time, the value of what can be done with data has done an excellent job of capturing the imagination of business leaders. The advent of AI hype supercharged this process. Avoiding the AI conversation amongst the senior ranks of organizations around the world, including governments, is practically impossible. With such voracious appetite for highly data-dependent AI it could be easy to assume that data and information management are enjoying similarly high profiles. This is not always the case.

Data quality, governance and broader data and information management are decades long-established areas of IT, with volumes of best practice, technical insights, and a broad and evolving set of software solutions available. What has not always been as available is the investment of resources to pay for these data-focused efforts. Without wishing to oversimplify, it’s generally a lot easier to get support and resources for the outcomes of data use, than it is the data inputs that make those outputs possible.

Asking The Right Questions To Provide Data Visibility For Everyone

Establishing visibility into the data used by AI solutions does not have to be purely technically focused. In my view, often the technical aspects of assessments are relatively well catered for. A pressing need exists to provide visibility for the less data savvy. This means thinking more about how data is gathered, its availability and its governance, and less about the technology that manages it.

The AI-TQ considers several areas of assessment under Source of Truth that look into data, its provenance and collection, its availability for inspection, and how it is managed and governed:

• Publicly available data - is the data freely available to anyone? E.g. Certain types of government published records.

• Proprietary data - is the data the private property of an organization? E.g. Customer Relationship Management (CRM) records.

• Auditability - can the source data be examined by anyone?

• Opt-in / Opt-out - is the data used gathered on an opt-in or opt-out basis?

• Reliance on Third Parties - to what extent does the solution rely on proprietary third party data and third party AI technologies, e.g. Large Language Models (LLMs) for GenAI from other vendors.

• Copyright - is the data subject to copyright? If so, has permission been granted for its use? What is the copyright status of outputs of the solution developed, is it clearly defined?

• Documented Data & Information Management Program - is the data of the assessing organization subject to a documented data and information management program that establishes governance, quality and use rules, etc.?

Coming Next To The AI-TQ: Transparency

This research is part of a series that will culminate in the official launch of the Artificial Intelligence Trust Quotient (AI-TQ) assessment. Next in the series is “Transparency” - shedding light on the technically opaque field of AI.

Also, my continued thanks and appreciation for the feedback and comments on this research so far! It is immensely valuable to me, and I look forward to more.

Making Values A Part Of Technology Assessment

The values of an organization are - for many - critical to successful outcomes for all their stakeholders, whether customers, employees, investors, or regulators. From personal experience, I have worked, and stopped working with organizations based on my perception of their commitment to their stated values. They are intensely subjective and even where the language used aligns between organizations, definitions and interpretation may not.

Someone outside of that organization, as in this case building an assessment tool, will find defining the values an organization adopts and what they mean in context is hard. Given this subjectivity, one-size-fits all use of values in an assessment is challenging, if not impossible, without making unwelcome judgements about those values.

Allowing for that subjectivity is, however, possible and it is a core feature of the Artificial Intelligence Trust Quotient (AI-TQ). Working from the intent behind including values in the assessment, the question that needs to addressed is simple:

“Whatever the outcome of other assessments of this technology solution, does it align with, sustain and enhance our organization’s values for all our stakeholders?”

Values Depend On The Organization

The AI-TQ does not attempt to impose any specific set of values on organizations using it. Doing so would be to assert the values of a third party. It does assert that values are an important part of any organization’s fabric and sets an expectation that they will be assessed as part of the process. As discussed, what those values are and their meaning is purely a judgment for the assessor. They could be solely grounded in financial metrics such as revenue and profit, or encompass far broader objectives such as minimizing climate impact, or promoting workplace equity.

By taking this approach, the AI-TQ aims to provide organizations the opportunity to ensure that their stated values are incorporated into its assessment of technology: without attempting prior judgment of those values.

Bringing Your Own Values To The AI-TQ

The AI-TQ allows for configuration to adapt the assessment to the organization using it. In the case of values, the assessment requires a minimum of three values statements to complete the section which are each scored against three criteria:

• Alignment - does this technology align with the value in question? Could it be seen as potentially damaging to this value?

• On-going Protection - does on-going use of the technology enforce the value over time? What is the potential for divergence?

• Extend - does this technology help share and promote the value? For example, amongst customers or suppliers.

While there is no maximum number of values that could be added, many organizations have a single digit number (often around five) of value statements. Generally keeping the number of value statements around this will offer the right balance of breadth and detail.

Coming Next To The AI-TQ: Source of Truth

This research is part of a series that will culminate in the official launch of the Artificial Intelligence Trust Quotient (AI-TQ) assessment. Next in the series is “Source of Truth” - a closer look at the fuel of AI, data.

Also, my thanks and appreciation for the feedback and comments on this research so far! It is immensely valuable to me, and I look forward to more.

There are many ways to assess the development, adoption and use of technology. Common approaches often include and combine business cases to justify investment, the necessity of regulatory compliance, and detailed technical capability assessments.

Artificial intelligence (AI) is different. AI is not only the current hype technology, but also one with far greater disruptive potential than those that have gone before. Incremental gains in productivity drive much of AI’s immediate value, yet it has the potential to fundamentally impact the day-to-day of people’s lives. From time and labor savings made with automations through the wholesale replacement of jobs and industries: the breadth and depth of AI-powered solutions’ reach is hard to ignore.

This is not AI doom-mongering. With major corporations and government organizations talking openly about how AI will disrupt current models of work and economic activity, it is time to expand the technology assessment tool set to account for the age of AI.

The Artificial Intelligence Trust Quotient: Bridging The Trust Gap

The Artificial Intelligence Trust Quotient (AI-TQ) is designed to help address some of the gaps AI technologies are exposing in existing assessments of technology. It is my view that the biggest gap being exposed is that of trust.

Trust, it is often said, is hard to build, easy to lose and difficult to rebuild when lost. Trust is also more important in the technology and software world than might first appear. An easy example is do you trust your bank’s financial software to keep an accurate record of your transactions? What would happen if you lost trust in that bank’s technology? Across many uses, different technologies require varying degrees of trust, the confidence in their ability to effectively do the task safely and securely.

The level of trust required of technology is highly variable, but in very broad terms, it depends on the negative consequences of something going wrong. So if your streaming TV service plays the wrong show, perhaps you’d be mildly annoyed but not very threatened. However, if an AI-powered solution is tasked with some role in your personal safety, the level of trust required is clearly far, far higher.

Given the vast range of uses AI is planned for and being put to, paired with its disruptive potential and a perspective on the necessity of trust starts to become clear.

If you accept the important role of trust in technology, and that particularly with AI technologies that a gap could exist, it begs the question. Do the existing assessments of necessary technical standards, required functionality, and compliance with regulatory requirements address the problem? I don’t believe they do.

An Assessment For Everyone, Not Just Experts

The AI-TQ’s purpose is to help people who are not technical experts establish a level of trust in an AI-powered solution. It does this by exploring the alignment of these technologies with the standards and values of their organization in easily understood terms.

Think of these standards and values as the means by which that organization can explain its use of an AI technology to the broadest group of its stakeholders. From customer, to employee, investor to partner. To do this with purely technical assessments involves a level of expert, technical understanding which the vast majority of people do not have. Relying on regulatory compliance suffers a similar problem; arguing that it doesn’t break any established regulation does not rule out misuse.

With AI technologies, and others, I suggest a simple test:

“If what your organization is doing with this technology was in the headlines of, say, the Financial Times, or Wall Street Journal, would you / your customers / your boss / your employees / your investors be comfortable with that?”

If there is any doubt how that question is answered, the AI-TQ is designed to help.

The AI-TQ has eight areas of assessment, each with a range of questions that explore some of the risks and issues that AI technologies potentially create. These are:

• Values - The values of an organization are an essential part of its mission or vision. In many cases they are central to guiding its actions whether from how it works with customers, develops products, or treats its workforce.

• Purpose - Within the AI-TQ assessment, the purpose of an AI-powered product covers two broad groups of use cases: commercial and civil.

• Source of Truth - Data is the fuel of AI-powered solutions and a major area of concern when it comes to privacy and security. What data is used and how, and on-going management and governance is critical.

• Transparency - It is undesirable for any technology which has the ability to impact outcomes for individuals, groups or organizations to be a “black box”, unexplainable to those subject to its decisions / outcomes.

• Human Governance - Human governance, like the AI-TQ itself, is designed to provide a human-centric perspective on the often highly technical capabilities which AI solutions are built with and upon.

• Outcome - Different outcomes and decisions have different magnitudes of impact, affect different groups or categories, and may, or may not be automated.

• Personal Characteristics - Personal characteristics found in training and other source data represent a significant risk of bias and toxicity for all AI-powered solutions.

• Regulation - Regulation of technology and software is not new, but regulation of the emerging capabilities of AI is a very active and rapidly evolving conversation.

The AI-TQ is not designed to replace technical and compliance assessments but to supplement them by creating a broader understanding of the solution, how it will be used, and its impact. Creating this broadly shared understanding is key to generating trust in the technology. Rather than being a magical “black box” the technology becomes explainable, its decisions auditable, its source data identified and understood, and its purpose clear.

Pilot Research will be publishing content that defines and explains each of the AI-TQ’s assessment areas over the coming weeks. Naturally, I’ll welcome your comments and feedback on this work, not least because openness and collaboration are, in my opinion, critical to how we choose to develop and use AI-powered technologies.