These are gorgeous, thank you for sharing.

If AI extruded text is not as good as human extruded text (derogatory wave to "pastoral counseling" and "conversion therapy"), then we can easily prove it, let's run the test!

If you still object, isn't your objection to machines themselves, and your concern isn't the quality of the therapy?

How would you characterize peer-group counseling and life coaching, vs. an AI chatbot?

None involve a professional therapist, nor are they practically distinguishable by the recipient, and yet you loathe one of them.

How is it barbarism?

(I did not see that coming and am curious.)

I bet people supporting AI therapy are willing to be held personally accountable when the systems fail...

.... at exactly the same rate as people opposed to AI therapy are willing to be held personally accountable when they prevent intervention that could have helped.

I hear you saying it's a prototype until it replaces a larger percentage of the existing system?

Urban driving is 70% of US driving and Waymo has 100 million miles. What exactly do you anticipate Waymo learns in the next city, which invalidates the results from the current cities? What changes?

I don't agree. This is system deployed in multiple cities, driving autonomously in real-world conditions, for over 100 million miles, and significantly outperforming human drivers.

What would you need them to do in order to demonstrate success?

I agree highway driving is the most dangerous, but high-speed driving is somewhere between 25-40% of US driving miles (as far as I can tell), we can ignore that and still have a pretty big win.

I agree Waymo has not released (all) of the relevant sensor data, but they've released trip metrics, which is more than "analysis".

Unless they're full-on lying about that data, I think we're confident without sensor data?

This specific piece is clickbait.

A (1) "philosophy professor", (2) cites unnamed sources and (3) anecdotal reports from "colleagues".

And how would you measure this, anyway? The hard question is if this is skill atrophy in the same way forging arrowheads atrophied our flint chipping skills.

Right! That's what I thought.

Headline references boat bombings, so not citizens or on US soil?

I'm confused. Isn't he incorrrect, I was under the impression that the due process clause can only apply to persons with a substantive connection to the US?

I agree, there are international rules violations, but not "due process"?

AI being used in writing research papers (AI can’t yet reliably write a research paper on its own), seems fine as long as there’s detailed human review/oversight?

Peer review (and professional conference organization) has always been a scam: labor donated by academics (so paid for by research funding), so commercial entities can then publish the research and sell it.

It can’t die soon enough.

I don't understand how AI (coupled with robotics) doesn't eat all the jobs in the next 50 years.

A rational government would be figuring how to translate that inevitability into UBI, among other things.

Instead, we have this (waves hands vaguely).

Is it not possible to say "you can be here if you want, that is, there are success stories if it's what you want"?

Curious why this means drawing a line on not doing those kinds of tech talks.