Its a shorter article this week as I am off to Lueven in Belgium for the kick off meeting of a new project for schools in Europe, AI Ready. I'll report on that meeting next week.

But as I write, the news is buzzing rumours about a market correction, or put more vividly a burst bubble, in the value of AI companies. Its been some time coming. Market prices of companies like OpenAI, Microsoft, Anthropic and of course Nvidia have soared in recent months. Vast sums have been promised for developing the AI infrastructure with limited. justification. Indeed much of the money sloshing around in the AI bubble is the major AI companies lending money to each other in a merry go round fashion. And whilst the GenAI providers have many individual users, there is limited evidence of mass take up by enterprises and less so of actual productivity gains facilitated by AI adoption. AI agents have failed to convince and Vibe programming appears to require much human effort in correcting bad code. And perhaps most worrying for the Gen AI providers (and for these investors) there appears to be limitations on the scope for major innovation through scaling Large Labour Models, indeed recent AI discussions on the non tech forms seem to be about AI shopping and AI slop (with AI produced music dominating the charts).

So what is likely to happen next. AI veterans are scarred by the memories of previous AI winters when funding dried up after periods of investment and excitement. And course this is common after previous bubbles, for instance in railways in the late 1800s. My view is that most of the big companies will get through with some scaling back although it is hard to understand how OpenAI can keep up its promised development programme, given it present relatively limited income. Its also hard to see how much investment will really result in new infrastructure, but then again its worth thinking about how much extra capacity is really needed. In this case, one of the winners may be Gen AI enterprises based in China which is developing an impressive leadership in cheap renewable energy.

Most national and may regional governments have been keen to see the emergence of start up companies in the AI business and I wonder how many of these will survive the oncoming 'market correction'.

Another big question is whether or not Large Language Models based on machine learning is really the right way to be going for future AI. American AI developer and commentator, Gary Marcus has long been critical of the Gen AI boom called for more investment in symbolic AI and the development of world models.

But even if there is a sharp fall in market confidence, I don't expect Gen AI to go away. Its just perhaps that we need to add to AI Literacy an understanding of how capitalist market act irrationally when faced with innovation. And hopefully see more development of AI for education based on Open Source Language Models and utlising the technical and cross disciplinary skills of staff from inside the education sector.

About the Image

Digital collage depicting servers extracting water from a local community, symbolizing how data center operations contribute to erosion, water scarcity, and drought.