Luna did not reach the decision entirely unprompted. Conversation logs between the lab and Luna show that the agent had created an attendance policy but later lost track of it, allowing the employee’s lateness to continue for months. Andon Labs eventually asked Luna to search her memory for its policies and assess whether the worker was still a good fit.

Luna agreed that the employee should be let go.

Petersson added. “But what we’ve seen in this experiment is not that the AI would be more ruthless or be worse for the employee in that decision.”

No, what we’ve seen is the LLM not actually doing the managing, and just agreeing with a human prompting it.

LLMs are excellent at agreeing with the people promting them.

I have no idea whether $100k is a sensible budget for starting a store, but they note that it isn’t profitable.

Batteries are far heavier. There’s going to be as many motors are there are propellers, it really doesn’t make sense to do otherwise.

It breaks convention to solve an entirely nonexistent problem.

I don’t think we’re going to agree on any points whatsoever, I’m afraid.

I think their method is pointless, that introducing new formats that are indistinguishable from well established ones is actively harmful, that their reasoning is flawed and raises mainly irrelevant points, and their only valid point is that fractions with denominator 256 are stored exactly, but then if the whole point of converting your colours to floats is to do interesting transformations of the colour space that you can’t do with integers, that’s all gone as soon as you start manipulating the floats, and all your carefully designed exactness went down the pan. If you want exact, keep it to the integers 0-255.

Neither[-0.5/255) nor [0,0.5/255) are some sort of disadvantaged and discriminated against group of humans that need more representation and I don’t understand why on earth anyone would care even a tiny little bit if they were under represented nor why you would describe it as unfair.

The whole thing is really, genuinely deficient in benefits and has obvious drawbacks.

I’m all out of entertaining things to say! Have a great week.

Yes, indeed. If you’re randomly generating colours and you want to evenly cover the {0,1,…,255} integer space, then generate a random integer in that range directly, or truncate a random float in [0,256). Generating a random float in [0,1) and converting it afterwards, then complaining you’re not getting enough 0 or 255 is silly.

This is a stupid debate. Of course you should use 255. Everybody uses 255.

The standard algorithm means that if you go from int(0-255) to float, black stays black and white stays white. The silly new one turns the extremes very slightly grey.

This needs stomping on, because inventing a conflicting standard where there’s currently only one is really really stupid.

If you actually want an even distribution of random colours (who on earth wants that?) then instead of generating random floats, generate random int(0-255). It’s not hard. It really isn’t.

I bet you that float values that round to 0 come up in float colour data way, way, way more frequently than values that round to 0.5/256, nullifying the silly mathematical point that you might get 0 less often. No. No you won’t. Not in real data.

Yes, it did, and that’s what I suspected, but I saw other possible interpretations and ironically, I didn’t want to assume. Sorry for being annoying. I only meant any of this for fun and didn’t mean to offend or frustrate you.

I was hoping your reply would be something like “am woman” in which case I was planning to tell you that in real life it feels like a majority of the women that I thought were really really cool but who I didn’t feel inclined towards romantically turned out to be uninterested in my entire gender romantically, so I was going to assume that you are also really cool even though my maths teacher was very insistent that correlation does not imply causation.

I assert that your “I have good news for your wife, then!” answer fits with my correlation.

If your reply was “you’re really rubbish at flirting” I was planning on responding with “everybody wins”.

I was hoping to come up with something amusingly self deprecating if it was along the lines of “I’m specifically not interested in you”, but I hadn’t got as far as thinking through that answer, hoping it wasn’t that.

As you can see I’m not all that great at this social interaction thing. Genuinely sorry to have annoyed you.

…which is?

Well, I like yours too, but don’t remind my wife of that. She’s just not a fan of me flirting with other men.

After all, why wouldn’t Microsoft use copilot to wrote their press releases?

"To address this class of risk, we use a defense-in-depth strategy with safeguards that block malicious instructions at multiple points and help keep tasks aligned with users’ requests,” Microsoft said.

Keep talks aligned with users’ requests?!? We’re doomed!

“We encourage customers to install the latest updates, use multiple layers of security protection, treat content from unknown sources with caution, and review AI-generated content before using or sharing it.”

No, no you don’t, you encourage customers to spend more to give copilot access to all of SharePoint and all of Exchange, so they can replace human competency, human expertise and long standing employees with copilot until they’re dependent on your ever-pricier subscription and you can raise prices until you recover your vast and incomprehensible LLM losses from your customers.

“Separating instructions from data may be part of the solution, but I think the distinction between data and instructions is not always clear in real-world workflows. For example, a user may ask an agent to arrange a business trip, requiring the agent to retrieve an email specifying the approved itinerary and a document containing the booking procedure,” Måløy [vulnerability researcher and discloser] said.

Or you could hire a PA, who wouldn’t spread the already live copilot worm.

A bunch of security experts note that distinguishing between instructions and data is a solved problem with SQL injection attacks, but completely unsolved in LLMs:

“None of them are rewarding that work commercially right now, so treat that as a multi-year research problem, not something a CISO should wait on.”

Is this a minor technical issue or a major problem? It’s a major problem.

Mike Wilkes, enterprise CISO at Aikido Security, said it would be difficult to overstate the potential problems from this situation.

That is actually pretty interesting.

I feel seen.
I’m still a tcsh guy.
My trusty .tcshrc still has some aliases that have been their since the 90s.

I support this amendment.

Came here for this.

tcsh is the bees knees.

Just as well trains and other vehicles don’t have roofs then.

Fun fact: Unlike natural gas, which needs enough oxygen and enough gas to explode, hydrogen is explosive at almost any concentration of hydrogen and oxygen!

Boom! Woohoo!

No. No it isn’t. That’s not how it’s made and not how it’s used. Its overwhelmingly made by the fossil fuel industry and calling it green in any sense is just greenwashing mostly the worst environmental problem generators the world knows.

Tokens aren’t going to get cheaper. Tokens must get more expensive, and soon. AI companies are making big losses even if you ignore the stratospheric debt for the immemse quantity of hardware.

they haven’t actively and personally killed people

“just following orders” … “they’re just customers, its not our business to judge”… “bottom line is the bottom line” … “I’m just a clerk, they’ll do it whether I’m processing the paperwork or someone else does” … “I’m sure that if we didn’t provide the software that enables ICE to target and tear apart families in a clear and flagrant breach of the American constitution, someone else would, so really it doesn’t make any difference if it’s us profiting from illegal detention and mistreatment or someone else” … “Israel are our allies and it’s good that American companies have international contracts”

Plus you absolutely do not want the entireity of the whole nation’s health data provided (not just free of charge but for negative cash) to a company with such disregard for privacy that I’ll be not just surprised but shocked if they don’t sell it en masse to insurance companies and anyone with deep enough pockets for it to be worth their while. And they’re using it with AI as part of the contract. Guess who’s going to be selling healthcare AI based off a whole population’s data provided to them at negative cost? Guess, go on.

No, Palantir is an evil, evil, evil company, the IT suppliers to fascists and it’s naive in the extreme for you to excuse all this because they don’t get their hands dirty on actual physical immigrants and someone else pulls the trigger based on their targeting data in Gaza.

They are ideologically opposed to the NHS and it’s utterly braindead stupid to hand over control of the NHS systems to people who want to destroy it.