Couple AI’s self-preservation and hacking abilities with the following news story this week is a recipe for disaster:
XPeng’s Milestone: The Chinese electric vehicle and robotics maker commissioned a specialized production facility where advanced humanoid robots are assembled and autonomously walk off the line. [1] (Instagram), [2] (XPENG IRON Humanoid Robot Now Walks Off the Production Line)
Asking chatgpt to analyze you just from your username on swissforum ist interesting…
no. its totally useless
You are welcome, but forgive me if I disagree. You have no {expletive deleted} what you are talking about.
True
Some similar views
Trump should listen more closely to his voters.
Compared to a year ago…AI has improved the results for queries about Python code. I can now copy/paste AI results into local development GUI and it runs without error warnings.
I make punctual queries, so the results are 10-20 lines of code. I still don’t trust more complex answers.
Have you tried the coding harnesses? e.g. Claude Code? I let it write substantial amounts of code.
I also connected, search and my files so sometimes ask it to research XYZ and save the results in XYZ file. e.g. recently asked it to “read XYZ web page of japanese restaurant recommendations, check locations against my itinerary in XYZ file and then put recommendations where they fit my itinerary into the file and validate the details by web search”. a one-shot request.
Paul Krugman shared his views why Trump needs AI.
Robert Reich has a good post on the subject, urging us to follow the money: As he notes, members of the Trump family have personal financial stakes in the industry. And one motive one should never, ever dismiss in current U.S. affairs is Trump’s personal greed.
I would, however, add two more points.
First, I think Reich is too glib in dismissing the idea that Trump imagines that he knows what he’s doing. “It can’t be,” Reich writes, “that he knows very much about AI.” But does he know that he doesn’t know? That Truth Social post above, in which Trump declares himself perfectly able to manage this wild new technology because he is a “STRONG AND SMART (High IQ!) PRESIDENT”, is like a textbook illustration of the Dunning-Kruger effect, in which incompetent people are highly confident in their own judgment because they’re too incompetent to realize that they’re incompetent.
Second, it’s important to realize how much Trump needs an AI boom.
Trump came into office believing that his tariffs would produce a huge boom in “manly” jobs. In fact, job growth has been weak — and more than all of the job gains have gone to women. Meanwhile, tariffs have driven up consumer prices.
And then there’s the Iran war, which has been an economic as well as a strategic disaster. Diesel prices are now well above $6 a gallon. Trump officials love to say that gasoline was over $5 a gallon under Biden, which was true — for one week .
Yet one thing did seem to be going right under Trump II. Enthusiasm for AI was generating a huge investment boom and supporting stock prices. Trump officials, notably Scott Bessent, the Treasury secretary — who Edward Luce of the Financial Times now dubs “the Pete Hegseth of the Treasury” (ouch) — had taken to claiming that AI will solve all their problems, ending inflation by reducing costs and curing the budget deficit by generating huge economic growth.
The backlash against AI, then, is taking away the one thing the Trumpists thought they had going for them. So in a way it’s understandable that Trump is trying desperately to hold on to the magic (and possibly endangering the future of humanity, but he doesn’t care about that.)
But of course it won’t work. The AI boom, once Trump’s last remaining political asset, is now a huge liability.
Henry saw it already years ago.
Now this could be something really useful, as AI “hears” far more inputs than your average psychologist/psychiatrist!
https://knowablemagazine.org/content/article/mind/2026/how-ai-can-listen-for-signs-of-schizophrenia
So long as the AI can avoid being infected
Look up " shared psychotic disorder"
In these rare cases, a person with an established psychotic illness passes specific delusional beliefs to a close family member or partner who is socially isolated with them
I think that an AI psychiatrist could be a future job…
Think about it, higher intelligence seems to come with a baggage; as in all colours of mental disorders.
I would say that at lesst a large part of those who create these things are on the spectrum, so why would their creation be unaffected?
Imagine an autistic AI with the morals of a seagull.
Or an AI like Startrek’s Landru who wishes to help humanity… at ALL costs.
By what I read about the topic makes me wonder if we would recognize a self aware AI if it did not want us to?
Wasn’t sure where to post this – this seems as good a topic as any:
When I started at Google a little over 20 years ago, lines of code written were still part of the (then still quarterly!) performance reviews. Pretty sure they’ve in the meantime reached or maybe even surpassed the tokens spent metric.
I made much money in industrial programming in my youth by finding and removing errors. Usually I stripped the assembler code by hundreds of lines while doing that.
However, I had to include some effects of those errors again because some workers got paid by those wrong metrics and would have gotten less money without doing anything wrong. Live was hard even before AI.
Oh, that became a metric at one point, too: lines removed.
Sir, I had to live fix your spelling.
Anyway, Charlie Munger summarized it best when it comes to metrics:
“Show me the incentive and I will show you the outcome.”
Thank you very much! I will find an amazing use for AI if it does that for me. Please keep helping me until then!
