All of the media is abuzz with skepticism about the potential returns on investment from the massive spending on generative AI by big tech companies. There are concerns about large investments on the one hand and uncertain timelines on the other. Analysts are emphasizing the risks of overbuilding infrastructure that the market may not be ready for.
Reputed names such as MIT’s Daron Acemoglu and Jim Covello of Goldman Sachs have struck a note of caution, with Acemoglu seeing a limited economic upside from AI and Covello arguing that the technology cannot solve the complex problems that would justify the costs. Sequoia Capital’s David Cahn has pointed out that the AI industry’s significant capital requirements—estimated as $600 billion annually just for hardware—may not be sustainable without more immediate and impactful results.
Is artificial intelligence really a damp squib then? If you have found yourself asking this question, it will be worthwhile to look at the framework of the “hype cycle” developed by the research and advisory firm, Gartner. It describes the way people tend to respond to new and emerging technologies. Among other things, it speaks of a movement from the “peak of inflated expectation” to “trough of disillusionment”.
The evolution of older technologies stands testimony to Gartner’s hype cycle. In 1946, when the Electronic Numerical Integrator and Computer (ENIAC) was presented, bold headlines were seen around the world, referred to it as an electronic brain, magic brain, wonder brain, wizard, and man-made robot brain. While stories about the ENIAC created short-term enthusiasm, it was followed quickly by disappointment as it did not seem to live up to its overhyped promise. Fast forward 80 years and the impact of computers in our lives today is all-pervasive. We have seen this pattern with other technologies as well over the last few decades – think railways or solar power, wind power, or even the internet before the dot com bubble burst.
So it has been with Generative AI. The media hype cycle, was in recent times, triggered by the release of ChatGPT 3.5 which fed into the media hysteria about the future of artificial intelligence —how powerful it will become, how quickly, and what it will do to jobs for all of humankind. The discourse was unhinged and it quickly devolved into a misinformation epidemic. Clickbaity headlines led to a distortion of the general understanding of what AI could and could not do, with typical hyperbole. In short, inaccurate and speculative stories about AI, fueled in part by charismatic AI startup CEOs, created unrealistic expectations for the field, and surprise surprise, the near-term reality didn’t meet the future predicted by the media frenzy. In other words, we first overestimated its speed and then in the “trough of disillusionment” are massively underestimating its magnitude. Needless to say this peak and trough is damaging for any technology. It leads people to believe that the technology is not the tsunami that it likely is, which can slow down adoption and attendant increases in productivity.
It is, however, important to remember that the way a typical hype cycle also works is that investments that accompany the “peak of inflated expectation” help in the building of infrastructure, leading eventually to its mainstream adoption. Little surprise then that experts like Joseph Briggs, Kash Rangan, and Eric Sheridan remain optimistic about AI’s economic potential and have been vocal about its ability to generate returns beyond its current phase. Research is already showing that organizations are seeing material benefits from GenAI use, reporting both cost decreases and revenue jumps in the business units deploying the technology. While McKinsey Research, 2023, estimated that generative AI could add the equivalent of $2.6 trillion to $4.4 trillion annually across its 63 use cases, McKinsey’s 2024 report explored the value created by gen AI use by business function. The function in which the largest share of respondents report seeing cost decreases is human resources, while respondents report meaningful revenue increases in supply chain and inventory management. Capgemini’s 2024 research, similarly evaluates the benefits that generative AI has brought at the organizational level in the past year, showing that on average, organizations realized a 7.8% improvement in productivity and a 6.7% improvement in customer engagement and satisfaction. Other examples abound.
Making real progress in AI will require both – time for adoption and realization of value in enterprises of all sizes, as well as a public discourse that is sober and informed. In the meantime, we neither need to make AI the subject of our utopian hopes nor our dystopian fears. In other words, just as it was a risky proposition to position AI as a panacea for all the ills plaguing society, it is equally risky to write off the value already being created by the huge investments in generative AI models and infrastructure, all of which lead to long term, tangible benefits.
In Gartner’s hype cycle, eventually, the reputation of a technology recovers and transitions into the “slope of enlightenment”, thereafter leading to the “plateau of productivity”. So it is likely to be with AI. The task at hand in the meantime is to ignore these hype cycles — both up and down — through responsible thinking and credible reporting- to ensure effectively “crossing the chasm”.
