A number of influential tech figures have been showing an uncharacteristic mindset when it comes to AI lately: healthy skepticism.
Salesforce CEO Marc Benioff said at a recent conference that some of his fellow tech CEOs had been “hypnotized” by the push to buy chips and build data centres, without thinking if it is really necessary.
Thomas Wolf, co-founder and chief science officer of machine learning library Hugging Face, pointed out that even the most advanced AI isn’t really that innovative — the way developers are currently approach it is just creating “yes men” repeating info, instead of systems that can actually think outside the box.
Chinese firm Manus created instant hype when it claimed to have created a fully autonomous AI agent. And while some people bought in, others put it under scrutiny, finding factual errors and an inability to do the most basic tasks.
Why it’s happening: One big reason is DeepSeek. The Chinese firm debuted AI that could match the leading models out there, but with a fraction of the budget and processing power. It was joined last week by a model from ecommerce giant Alibaba that further reduced computing needs.
That is subverting a belief that has driven AI investment: the only way for AI to improve and become capable of reasoning is by throwing more money at buying chips and computing power.
But also: People and businesses are not finding the skill or efficiency gains of AI to worth paying for (at least, not enough to generate a profit). And some people are getting a little tired of waiting for AI to start getting useful — and generate enough revenue to justify the investment.
“AI agents” capable of multi-step tasks and figuring out for themselves which apps to use in order to complete them have promised to close the usefulness gap, but the results so far have still been underwhelming.
Why it matters: For one, it is taking critiques from skeptics and writers off of social media, blog posts, and newsletters and putting them into the boardroom. But it is also striking at each of the economic pillars propping up the AI boom, and they could come falling down.
If the people in charge of the money stop spending big on chips or start questioning when their return on investment is going to come, massive tech companies that have driven stock market growth over the last year could see their stock price crater, while some big money firms may have to write off huge investments.
It may already be starting: An analysis last month showed that Microsoft is cancelling leases for new data centres, suggesting it may be rethinking how much computing demand there will be for AI going forward.
OpenAI has reportedly gone to Japanese investment firm SoftBank for a massive US$40 billion in funding. While some might look at that as fueling future growth, doing so just months after getting over $6.6 billion in funding doesn’t provide a lot of hope that it is close to changing the fact that it is burning through cash.