Fake AI 

Edited by Frederike Kaltheuner

Meatspace Press (2021)

Book release: 14/12/2021

This book is an intervention - 

Chapter 9

Uses (and abuses) of hype

By Gemma Milne

“Perhaps we need more boring AI”

Hype has always been a part of my career. As a science and technology writer and researcher I spend a lot of my time following the technology and science startup scene. I’ve been on the receiving end of bombastic pitches, I’ve helped people desperate to get noticed, and I wrote a book on how hype obscures the future. I believe that until we reckon with the dual nature of hype, snake oil and pseudoscience will persist in AI, and science and technology more generally.

Reactions to hype range from excitement at the genuine progress that AI has made, to scorn at lame sales pitches. The word has a variety of definitions too, depending on who you ask. Some see it as unfaithful to the truth, a lie of sorts. Others consider it as merely an ultra-subjective form of expression, a form of “fair play” marketing, as it were.

To me, hype is a tool for capturing attention. Hype is exaggerated publicity which inflates expectations and prompts emotions such as excitement or fear, which can stop people in their tracks. It’s a little like a magic show. By tugging on the emotions, hype causes people to lend their ears, eyes, and brain waves to ideas, claims, and sales pitches in an uncritical manner. The tricks are wondrous, but they’re entertainment.

The use of hype comes with a warning label though. Forget that it’s entertainment and it becomes lying, usually for nefarious ends. Furthermore, hype can result in accidental fooling. Out of context, hyped messages can very easily be misinterpreted. The originator may not have set out to trick or deceive, however, the fact remains that because of its spread, the language used, or the cultural context, hype can be misinterpreted as truth. 

Yet the role of hype is more significant than simply capturing attention at all costs. It eases the complicated, arduous, and expensive journeys between research lab, office meeting room or simply someone’s brain, out to the market and the public. It lubricates, bridges, and facilitates. Hype can be a strategic tool for selling ideas, products or political campaigns. It helps attract people, pennies, policies, and partnerships.

Hype is, in and of itself, a creator of value—regardless of the substance behind it.

Hype attracts attention, money, good employees, and social capital. Startups want the attention of investors. NGOs want to attract funding. Thought leaders want to lead thought. Researchers want better-funded labs. Universities want to be higher on the league tables. Companies want more clients. To align yourself with “innovative” narratives gives you a head over the competition.

That’s why hype is unlikely to disappear. It is genuinely useful for generating attention—a scarce resource in today’s attention economy. But where there’s opportunity, there’s opportunists, and so the existence of AI snake oil is unsurprising. Its prevalence, though, is dependent on the rest of us not reckoning with the problematic, dual nature of hype. This needs to happen on two levels.

First, those on the receiving end of hype need to collectively recognise the power, mechanics, and sheer volume of it. The sole purpose of a hyped-up message is to grab attention. So when those on the receiving end do not scrutinise the message further, a lack of context, or the absence of deeper understanding about the topic can easily result in accidental fooling. There are also places and circumstances where hype must be checked at the door entirely—otherwise the public, journalists, investors, and government ministers will continue to find separating snake oil from legitimate claims extremely difficult. Hype about technology, for instance, has no place in serious policy deliberations about how to regulate technology.

Enhanced critical thinking and a better understanding of how hype and marketing “work” can help reduce hype’s problematic impact. Its power, after all, is in its illusion, so seeing it for what it is can help stop it in its tracks.

Still, that’s not enough to stem or limit the volume of hype that is produced, especially in AI.

Secondly, those of us who work in science and technology—as builders, policy-makers, commentators, researchers, investors—must reckon with our own complicity in the use of hype. This requires introspection, acceptance, and possibly a pinch of low-level shame. Hype helps projects move forward. We might not use those words, but we know it to be true.

We could argue that hype is just a tool used on the road to greatness, that this lubricant is just a reality of getting through that complicated, arduous, and expensive idea-to-market journey. We can justify its wielding like advocates of placebos justify lying to unsuspecting patients. The cost of accidental fooling that hype comes with is worthwhile as it’s the only way to get stuff done in the dog-eat-dog world of the innovation industry. The ends justify the means when saving the world is at stake.

This pragmatic view neglects the dual nature of hype. The very lubricant that eases ideas through complex systems is what allows snake oil to make it to market. Hype is useful to capture attention in today’s attention economy, but snake oil is an inevitable by-product. Those who generate, amplify or simply cash in on the hype surrounding AI inevitably also create the conditions in which snake oil vendors and pseudoscience can thrive.

Lack of care characterises a lot of the problems around hype in science and technology. Those on the receiving end who do not listen might be ignorant, flippant or time-poor. Then there’s the fact that those creating the hype are often willingly complicit; carelessly wielding a tool which makes them look credible. If we reckon with these two forces, snake oil will start to have less success.

Beyond that, what’s needed is a genuine culture change in technology and science. The pressure to publish at all costs in order to progress in academia, the precarious working conditions in journalism, and the growing competition for funding are all systemic forces that create conditions in which problematic hype can thrive. This culture often equates value with attention, and links status and fame to trendy topics or cults of personality, rather than proof of positive social impact. Publish or perish should itself die.

Perhaps what we need is more boring AI. More cautious product pitches. More thought leaders who communicate uncertainty about their own thoughts.

Snake oil in AI will thrive until the various forms of hype that keep it trendy are deeply and unashamedly reckoned with by those who arguably stand to benefit most from no-one saying anything at all.

Gemma Milne is a Scottish science and technology writer and PhD researcher in Science & Technology Studies at University College London. Her debut book is Smoke & Mirrors: How Hype Obscures the Future and How to See Past It (2020).

Next: Chapter 10
Talking heads

by Crofton Black

Instagram        Twitter