dofm

Meanwhile, Meta is selling their spare AI data centre capacity.

If you are dragging within the page, then the draggable “ghost” is your problem too (it’s not automatic). As is deciding whether the ghost should affirmatively snap out of and into the source and target containers (that is, whether the destination container should appear to “make room” for the ghost at some transitional point), whether there’s a transition on that, etc.

Drag and drop in desktop non-web GUIs have all sorts of visual affordances that users expect for consistency that are your job in HTML.

I was once quite up-to-date with raw frontend work and I tried to implement this; it’s actually not trivial.

This is Jensen Huang we are talking about. One of the most significantly influential people on earth right now, one of the planet’s most obvious espionage and blackmail targets, someone who both companies and nations will be spying on, and he is shagging a foreign national — a Chinese citizen! — in secret.

Even if he has enlisted no company employees or resources in his deception of his wife, this is remarkably negligent behaviour on a professional level. Just enormous.

Be serious: this is a huge deal and they are for sure looking at his conduct.

Blackstone think that post-IPO Anthropic will start out with a junk credit rating.

FT seemingly covered it too if you can see it (I can’t):

https://www.ft.com/content/59383c7f-8d82-493f-bf01-d2e862a10...

A news organisation's liability for AI hallucination falls on the same person that the liability for printing a lie does: the publisher.

Architects and properly-designated publishers both have potentially long liability that is to an extent independent of their future employment. (The same is true of hospital consultants and others)

In significant cases, the way you provoke companies to do the right thing (security, sticking to the facts, not cutting corners in building plans) is to make someone legally responsible if they don't. That person's liability is limited by their actions — if they can demonstrate they discharge those responsibilities appropriately, they are in the clear.

That person takes on that designated role, does the job, and is rewarded properly for the responsibility it involves. They embody the push to reduce that risk.

So an average employee using the tool as directed by their employer, they do not have liability if the company has designated someone in this role. (And in this scenario, if they have not, there would be a punitive fine for that, independent of other outcomes).

I have no particular problem with the notion that a corporate AI safety/responsibility officer is a plausible way forward. If you are delegating responsibility to a machine, someone has to be on the hook. Without such things it's a wild west with lawyers. That's not sustainable.

(We're talking three decades ago now)

This is actually a pretty challenging standard, because the moment the CEO tries to gaslight their spouse/partner that it's not an affair, they are already at risk of bringing the company into a deception: working late is almost always the cover, for one thing. But at that moment you have a substantially compromised person as a CEO.

This might seem a silly place to draw the line, but consider what the escalations are — the CEO's PA being asked to mislead the spouse, inventing a meeting they were at, inventing an away trip, etc. etc.; they always mount up. There is no executive who has an affair who doesn't pretend they are at work, and there's no pretence that doesn't ultimately put an employee in a difficult position.

It's entirely reasonable for a board to say that if the CEO or the chairman is having an affair it is their business.

And this is before you even consider that the affair is ordinarily likely to be with an employee.

Now, if a CEO at Huang's level of geopolitical importance is having an affair with anyone, it's the government's business, and the shareholders', because it makes them a (local and international, commercial and state-level) blackmail and espionage target even if they are not literally boffing a spy, which in this case he evidently is.

So yeah, it matters.

It's satire.

Some can affordably deliver to Microsoft because they aggressively police what they are actually sending on their customers' behalf; others don't get involved and you end up having to spend a fortune on a dedicated IP.

The DMARC thing is just another part of this; one of my clients discovered this silent failure themselves a few weeks back and addressed this.

It's a bit like the idea of the additional responsibility/burden of the publisher of a newspaper.

You have to factor in all the people enjoying that the shitcoin-for-your-iris-scans guy is having his time in the comedy-drubbing barrel. Don't leave us out, we've been waiting a lot longer for this.

FWIW, "anodyne" is not the word I'd reach for when describing things Altman says — I much prefer "banal".

(Not sure you can classify me as a hater, at least not in my day job context. Cynical: it's a fair cop.)

We should stop using "solve" when we mean "releasing products to target". Nobody will ever solve this problem with LLMs as they are currently architected.

1) everyone can run into the lethal trifecta

2) the product cannot, on an architectural level, be fixed to stop it

3) they are helping the maker of the product go to IPO

If banks are not stupid, they'd say "hey, this presents an unacceptable risk to our customers and by extension, to us, maybe this isn't something we should be involved with".

But no.

In quite a lot of sectors, gambling on a prediction market for an outcome you can influence is frowned upon, unethical, even illegal.

To be clear, I am talking about the investment analysts who have a role in setting the IPO price, which is a (then-private) decision, not an outcome of a market. It would almost certainly be a crime for the analysts involved in pricing an IPO to make prediction market bets.

But this number isn't a matter of market volume, is it?

It's a matter of what analysts, regulators, banks, and investors choose.

And they appear to not be that keen on a valuation that high, judging by the press.

But they are still not likely to be investment-grade at IPO.

The most peculiar thing is that it expects to be able to "read" directories? Harmless, so it fails, but it's a waste of a tool call. I am increasingly wondering about modifying Pi's read tool so it returns a directory listing in that context.

I have had some difficulty getting it to have a nuanced understanding of how I want to manage git, but I am reaching the conclusion that it's me that needs to change there. Basically I don't think you can say "only commit when I ask you to" and then actually ask it to more than a couple of times in any session, because it will tend to then assume it can go ahead. Scaling the context up might help me a bit there.

But apart from this, I really like it.

I will test Qwen 3.8 27B properly at some point, (though judging by the insane, empty noise making campaign that is going on on social media, Qwen 4 is near)