After the madness of the crypto and metaverse hype, there was of course the suspicion from the start that the excitement around AI could also result in a bubble. But the voices of the doubters could hardly be heard. The screenshots of ChatGPT answers on social media were too dramatic. The chat self-experiments on your own PC or cell phone are too fascinating. Google, Meta and Microsoft as well as many other established companies were too determined to jump into the issue and invest billions upon billions. On the one hand, the media attention on the topic has of course waned somewhat. On the other hand, it must now be clarified what or whether the gigantic investments will bear any fruit at all. The Washington Post has put together a concise overview of why skepticism is appropriate. At least for those in a hurry, it should be required reading before any further discussions (for example, the Substack AI Snakoil by Arvind Narayanan and Sayash Kapoor – which I interviewed here) goes into more depth.
First of all: The now legendary number of 100 million users that ChatGPT was able to gain within three months of its launch is not really correct – it is an oblique interpretation in the spirit of Open AI. The number is seen as evidence that we are experiencing an extraordinary moment in technology history.
What’s missing are solid numbers: How much have Microsoft and Google actually earned from their generative AI offensive? Isn’t AI perhaps a billion dollar grave? The companies are just as opaque on this issue as they are on details about the exact functionality and training data of their language models.
Post reporter Gerrit de Vynck reminds us that the tech industry is littered with flops: dot-com, social media sites that no one wanted, crypto, metaverse, and the perennial favourite: self-driving cars
But it’s still unclear how and when this technology will actually become profitable — or if it ever will…“generative AI” is incredibly expensive to build and run — from specialized chips to data server computing power to expensive engineers.
The crucial question is probably: Can generative AI really be a quantum leap or just a gradual improvement of existing technology – because then it would simply be too expensive:
“At the end of the day, AI is just software, it’s expensive software,” Andrew Harrison, CEO of venture capital firm Section 32, said. “It’s low-margin software unless it does something that’s 10 times better.”