Some of you may have had fun with Chat GPT since it came out. It's a chatbot that can simulate, imitate and summarise human text - sometimes.
Now Sam Altman, its creator, wants $7 trillion for chips to make it smarter. Seven trillion dollars is the budget of the entire USA, a country with no statutory sick pay or maternity leave.
It's also 21 times the estimated cost of ending world hunger.
Is it sane to divert this much money to machines that, when faced with a picture of a clock face, will always tell you the time is 10.07, because watches are always photographed with the hands at 10.07, and that's all it knows?
It also told me, in response to 'Who is Sheila Hayman?', that I'd died in 2019 - which, if true, makes this article considerably more special than usual.
Why does it make these basic errors? Because it's a machine, which works in binary computer code by guessing the probability of the next bit of code, based on the unimaginable amounts of data on which it's been trained.
That's it. It 'knows' literally nothing.
Last week, in news that should have made bigger headlines, the chairman of BP announced that, far from reducing gas production, his company had been obliged to increase it because of demand from AI.
And language models like ChatGPT massively increase AI's energy hunger. If Google were to use chatbots for its nine billion daily searches, it would need 29.2 terawatt hours of power each year - equivalent to the annual electricity consumption of Ireland.
Meanwhile, another recent study discovered that training an AI by having it learn from what a four-year-old child sees could make it about 60% as clever as the child on 22 'visual concepts'. Twenty-two is a fairly small number: I think any of us with a hamster could probably think of 22 things it can recognise.
And what a child - or indeed a hamster - can do, but not even the biggest AI can, is understand what it sees and use that to build a model of the world.
How much energy does that child use? Around 20 watts a day, completely renewable - slightly more than fully charging a smartphone.
AI may take over the world - but it will fry it first.
- Sheila Hayman is a member of Climate Emergency Camden (CEC) climateemergencycamden.org.
Comments: Our rules
We want our comments to be a lively and valuable part of our community - a place where readers can debate and engage with the most important local issues. The ability to comment on our stories is a privilege, not a right, however, and that privilege may be withdrawn if it is abused or misused.
Please report any comments that break our rules.
Read the rules here