Do we value artificial intelligence too little or too much?
For now, artificial intelligence (AI) is certainly less intelligent than those who sell it to us say.
Think it's not worth what it already costs?
For starters, it doesn't give reliable information. If I typed in my name, say, again, now, in Chat GPT, I'd hallucinate again. The other day, for example, he said that I had as a pet... a chicken!
yourself
AI today tends to often make up data by extracting and cross-referencing it by mistake: Do you think you can trust AI to recommend a drug?... Don't!
And don't you think AI learns?
The answer is no. As it stands today, it neither learns nor will learn.
Why do other experts insist that it does learn every time you use it?
I don't think so, because being intelligent is not accumulating more and more data, as I already explained in La mente algebraica. For AI to be truly intelligent, it needs more genuine innovation than simply adding more capability to current models.
Is artificial intelligence dangerous?
What we have now is dangerously stupid, because what it does advance is in verisimilitude, as reported by NewsGuard, which fights misinformation on the networks and warns that Chat GPT 4 is as wrong as 3.5, but it is more frightening because believable in his mistakes.
What errors?
The other day, for example, Chat GPT was mind-blowing: he was giving impossible names to replace McCarthy as Speaker of the US House of Representatives, like Liz Cheney...who isn't even running!
Isn't Chat GPT a threat to Google to replace it as a search engine?
In this too, the AI is still stupid and that is the danger. Because what Google and Microsoft should be concerned about is not that I do Internet searches better than them...
Isn't Chat GTP replacing them?
No! What is really worrying is that Chat GPT fills the internet with errors and such hallucinations and ends up making any search unreliable to the point where people stop using it.
And that everything ends up being a lie?
It is what is called the echo chamber effect: the autophagy of AI and the great threat to searchers is that they end up intoxicated with their lies. And in the end, there is no way to know what is true or false when you look something up on the internet.
Is ChatGPT being a good business?
Satya Nadella, the CEO of Microsoft, admitted before a jury a few days ago that, after the surprise effect of the first weeks of Chat GPT, he did not stop using Google, but gradually deflated the bubble without that his Bing won the market.
So what is AI for?
You still can't drive self-driving cars... or only in San Francisco on some streets and at some times.
And to search for restaurants, hotels, plan holidays and visits, schedule...?
Google Duplex wanted to do the same a few years ago...
I don't remember, really.
Because it wasn't useful and we all forgot about it. What if in the end all this noise about AI just ends up making you a restaurant reservation?
Doesn't it at least help you do your schoolwork?
It depends on the teacher's requirement.
And will this money that is being paid in the stock market for the future of AI be lost?
We'll see: maybe there's a $5 billion a year market for AI; but not 100,000, which are played now. For them to be worth it, a disruptive innovation would be needed to make it reliable and efficient. And no one knows, not even me, if it won't happen in the end.
Do I invest in chips or not?
Remember before the internet was profitable there was a dotcom bubble and thousands of people went broke.
Will there be a China-US chip war?
There is already one and it is real. What is not so real is the assumption on which it is based, that whoever makes the most powerful chips for AI will conquer space and its resources to dominate the Earth.
He won't even give us an optimistic note?
AI has the potential to change the world in ways we can't even imagine: understand the brain, stop global warming, cure diseases or take care of the elderly with almost human robots...
Thanks for encouraging us.
But I must add that on this path AI can destroy democracy and cause a technological and scientific regression. And progress or regress depends on the decisions we make now.
