An AI crash is coming: don’t let it drag you down

All the signs are there that an AI crash is coming: overinflated claims and dodgy hype, a serious lack of expectation management, fears around privacy, ethics and fairness, over-enthusiastic amateurs misunderstanding complex and often opaque generative AI tools, whether that’s chatbots or picture creators, and of course the inevitable scammers.

However, this AI crash will not be a financial or technological collapse – another AI winter. Instead, its primary damage will be to AI’s public credibility.

We are already seeing the first hints of this in our research. When we surveyed a group of CIOs recently, 82% said that more than half of the vendors they encountered claimed some AI capability, yet in only 20% of cases did the claim appear believable (view our new infographic “But is that really Artificial Intelligence?” to learn more).

Hype easily leads to disillusionment

For the wider public, it’s not just the inability of many chatbots to sort truth from fiction and reference their blithe assertions. There is also the growing menace of maliciously-generated deepfake videos, while voice generation is now good enough that phone scammers have been able to cheat families into believing that a relative was in trouble and needed money.

Stepping back for a moment, a credibility crash makes sense: if generative AI is a technology that’s moving faster up the ‘hype curve’ than any technology before it, it’s also going to tumble into disillusionment a lot faster too. And even though AI researchers know that the large language models (LLMs) behind generative AI are just one corner of one segment of the entire spread of AI technologies, that public disillusionment is likely to infect more broadly.

Forget the chatbots and learn to use the tools

Yet amidst the concerns and disillusionment, AI as a whole really can bring value to the table. Generative AI has its limitations, but smart companies are already combining it with other technologies to create valuable results. For example, feeding the output from ‘traditional’ predictive AI or computer vision systems into an appropriately trained and constrained LLM to dig out hidden detail and produce reader-friendly reports. It helps too that it’s not just start-ups doing stuff like this – based on very recent conversations with IBM and Salesforce, among others, I’d say it’s all sorts.

For technologists and business leaders responsible for AI strategy, it’s essential to embrace this more nuanced and transformative view. AI extends far beyond the headline-grabbing chatbots to a wide range of other technologies and applications, and as part of this bigger toolkit, generative AI can add real business value. Don’t let the hype scare you, but don’t let it carry you away either.

Click here for more posts from this author

Bryan Betts is sadly no longing with us. He worked as an analyst at Freeform Dynamics between July 2016 and February 2024, when he tragically passed away following an unexpected illness. We are proud to continue to host Bryan’s work as a tribute to his great contribution to the IT industry.