radical Insights.

Weekly Research and Commentary on the Future of Business and Technology.

Three Arguments for Why This Hype Cycle is Different.

Sep 12, 2023

Last month, I met with an executive MBA group at the tail-end of their immersion week in Silicon Valley. They had made the site visits, leaned into the expert keynotes, seen the demos, and probably taken copious notes across the week to a steady drumbeat of bold (and bolder) claims about the AI revolution. When I joined them, they were duly impressed with the potential but had survived the week with a sense of pragmatic skepticism still intact. They were asking important questions and put a few to me directly in our conversation about tech trends, possible futures, and real opportunities.

Having witnessed the myriad Silicon Valley hype cycles of the last decade – including the most recent “nothing to see here” mass pivot away from Web3/blockchain and metaverse projects, many of them wanted to know: Why exactly should this super-hyped AI wave be different?

I see three pretty persuasive arguments to be made here. The first and most straightforward is simply an extrapolation of trends. As Mustafa Suleyman (co-founder of Deepmind) put it in a recent interview: “The scale of these models has grown by an order of magnitude – that is 10X – every single year for the last 10 years. And we’re on a trajectory over the next five years to increase by 10X every year going forward, and that’s very, very predictable.” At each new level of scale, the models have gained (sometimes surprising) new capabilities, and the history of digital technology suggests that all of those capabilities will eventually be made not only more cheaply available but also easier to apply to a given business context.

Speaking of context… it matters, and the second argument concerns applicability across contexts. Several of the recent hype cycles have fizzled out largely because the really cool things – while really cool! – didn’t meaningfully address real problems in the real world in a way that obviated or even surpassed existing tools and systems. This was glaringly the case for a whole raft of blockchain-based “solutions,” and while I’m still optimistic about the upside of industrial metaverse projects, most of the hype around most of the metaverse technologies didn’t actually translate into much real world value outside of gaming. Gaming, to be sure, is a huge market, but it’s not a huge market for a huge number of companies across a wide range of industries.

One thing the vast majority of businesses have in common: their key opportunities and addressable challenges exist in domains not easily reached by the tools of the metaverse. By contrast, the current and near-future crop of AI tools potentially constitute something like a general purpose technology with tremendously broad application and potential upside for companies whether they’re looking for disruptive innovation (most aren’t) or simply incremental innovations and efficiency gains (all are).

Anecdotally, I know very few people who have attempted to incorporate Gen AI / LLM-based tools into their own work flows and NOT experienced significant improvements in efficiency, breadth of capability, or divergent thinking that feeds into a larger creative process. And in our thoroughly digitized world of work – with all of its attendant pushing of pixels and manipulation of bits, almost every business is made up of individuals who could reimagine their personal workflows; just as many businesses (and eager executives) might reimagine their operations, workforce, and resource allocations.

The last argument I’ll cover here is one that borrows an idea from biology – specifically of an “obligate” behavior or feature as something necessary for survival. AI might become an obligate technology for firms (and even nations) in a highly competitive evolutionary landscape. Consider for a moment that big tech companies are already operating under this assumption. Project forward in time, and we might expect this behavior to ripple outward such that many firms will pursue potential AI-conferred advantages under the assumption that their competitors will all do the same to survive. This is a distinct dynamic with its own propulsive logic, and significantly, it’s one that we did not see as a pervasive part of previous hype cycles. Moreover, that logic feeds still more investment and interest into the factors addressed in the previous arguments here (around development and broad applicability).

So will this time and this tech be different? I think the answer is a strong yes – and different in ways that we’re only beginning to perceive. (via Jeffrey)