d1sxeyes

For anyone not aware, this was the Bahrain Grand Prix held in Malaysia. The FIA allowed Bahrain to host their Grand Prix at a circuit which isn’t normally used for F1 any more in Malaysia because of the geopolitical situation in countries around Bahrain.

Therefore this was the Bahrain F1, it just happened to take place in Malaysia.

More likely we will finally settle on something like “it isn’t conscious because only humans can be conscious”.

In the past, we’ve treated even other humans poorly because we think they’re different in meaningful ways, I sadly don’t see how this will be different with AI.

Well to be fair the output you get has improved greatly. You can still get billions and billions of Luna/Sonnet tokens within your subscription comfortably (doesn’t feel fair to compare Luna to Haiku). Sol/Astra/Fable… yeah, they’ll chew through your credit.

I think the legal argument would be that the owner of the recording device (in this case the TV) was informed and therefore responsible for informing others.

The question I think would centre on two things: is burying the information in a ToS acceptable for “informing” and is the amount of data collection proportionate for providing the service. I think the answers to both those is no.

It is however extraordinary, and there don't seem to be any reasonable measures a carrier could take to work around it.

Almost certainly not. The legislation says: "An operating air carrier shall not be obliged to pay compensation [...], if it can prove that the cancellation is caused by extraordinary circumstances which could not have been avoided even if all reasonable measures had been taken."

That gets them off the hook for even second order effects. No harm throwing in a claim though, there's no penalty for having a claim turned down.

Yeah but I don't think it's that legislation you'd rely on: you paid for a service they didn't deliver.

Compensation yeah, only applies if the reason for the delay/cancellation was under the airline's control. Nats down is pretty clearly a reason to reject compensation claims.

It’s clearly unreasonable to assume that LG can know and procure consent from everyone you choose to allow into your living room.

It’s reasonable to push the burden of procuring that consent onto whoever controls entry to the space, with terms like this.

As a thought experiment, let’s take a service like Zoom where you can record video of participants. It’s reasonable to say “you’re operating the camera, if someone appears in the background, it’s your responsibility to make sure they’re informed and consent to the recording”. In fact, trying to establish the identity of the person in the background for the purposes of procuring consent is probably a greater intrusion on privacy than not.

The problem here is different. The problem is that the volume of data being collected is clearly unnecessary for achieving the purposes it’s intended for.

Some third party harnesses allow you to do this, definitely makes sense.

That's not quite the case in common law countries. In countries like the UK and the US, the interpretation of the law as written is ultimately decided by the court, meaning that the definition of what constitutes a crime/felony/misdemeanor is subjective. The more courts that attempt to interpret that specific law, the more 'jurisprudence' builds up, meaning that the interpretation applied in a previous decision can certainly have an impact on a future decision.

Well, people with epilepsy who “prefer” the autistic label are just people who are wrong. They may be both autistic and have epilepsy. Epilepsy is diagnosable based on observed neurological events and physiology, it does not depend on behavioural observations, a person’s opinion about whether they have it, or a doctor’s interpretation of a self-evaluation (as autism diagnosis often does).

I disagree with you that it’s the discussion that’s off, it is the science.

The problem is that autism is not a clearly defined disorder. Its definition and diagnostic criteria can and have changed, and even then, those criteria are poorly specified.

One thing that I have always found troubling is the “clinically significant impairment” requirement of an autism diagnosis. It seems to me that a person’s level of “impairment” is likely to depend on individual differences and possibly environment and circumstances. To me this seems odd: surely the disorder exists regardless of a person’s ability to manage it? If a person grows up in accommodating circumstances, do they not have autism? What if someone grows up in tough circumstances but later finds more accommodating circumstances, is their autism “cured”? Questions are facetious, but the point stands.

It’s also odd that we see autism so highly implicated in comorbidities with other disorders such as epilepsy, which aren’t dependent exclusively on phenotype observations to diagnose. To me, this strongly indicates that what we call autism today (or some sub/superset of it) likely will become definitively diagnosable in future with improved understanding and brain imaging.

For now, it feels like the diagnostic criteria are actually just guidelines for guessing whether someone is autistic, rather than determining it (especially when it comes to determining what constitutes “significant clinical impairment” and its varying degrees).

To be clear: I definitely believe that “autism” exists and is a disorder that impacts people broadly as described, the challenges are around its definition and diagnosis. I have spoken to autistic people before who have found my views to be offensive, and invalidating of them. That’s not my intention, my intention is to pursue accuracy because I believe with better understanding, we will be able to make sure our public spaces/workplaces/etc. are accommodating to everyone, including folks with different needs.

No harm done :)

I’m interested, in what cases would it not be true if we assume “average” here to be “median”? I can only think of one: if we consider “understanding” to be discrete rather than continuous.

OP said: “You'll probably get lung cancer anyway if you continue it long enough”.

That’s not true, and is why I focused on lung cancer. Don’t forget that smokers don’t just die of cancer, so even if we take your source, we still don’t have the proper picture.

That’s why I also added the point that about half of smokers die prematurely due to their smoking, because I was not planning to exhaustively list every possible cause of death which could be caused by smoking.

It was a joke, you are, of course, technically correct, there’s no requirement for the mean and median to be the same.

That’s not true. The lifetime risk of a smoker getting lung cancer is about 10-20%, not “probably” (somewhere between 15 and 30 times higher risk than a non-smoker).

In fact, only about half of smokers die prematurely from their smoking.

I mean, I’m not saying start smoking, but also let’s at least have accurate facts.

Half of all Americans have lower than average understanding of statistics (or at least, lower understanding than the average American)

This reflects my own experience too. Generally speaking, incompetent people don't get promoted in the same way you normally understand promotion. They don't get their boss's job. They don't stay in their boss's boss's reporting line.

They move to a nominally higher level role in a different department, often a cost centre, generally to some role which serves the organisation just by existing (e.g. a legally required role, or something that serves some certification purpose, or so that sales teams can say 'we have a dedicated person in our X office for doing that').

In principle I agree but in practice I don’t.

The majority of high severity vulnerabilities are not the kind of thing you need a PhD in Comp Sci to comprehend, they are mostly about finding a way to get a system to end up in a state different than was anticipated when entering a particular code path.

Exhaustively looking at code and identifying ways to do this is something LLMs are quite good at. They don’t get tired, and you can run them non-stop.

They're also (generally) quite good at reading the literal meaning of the code, whereas humans often see the intended meaning first, and can be biased.

If you had a tireless junior engineer who was given the job of “make this application get into a state it’s not supposed to be in”, you’d probably get similar results.

What Mythos is quite good at is both the first bit and coming up with ways it could chain that together with other bits of unexpected state to create something that forms a meaningful vulnerability rather than a dead end.

It’s very easy, just change each third word into something different than what you primarily wrote.

It’s very simple, just change every third word for something different to what you originally wrote.

It’s very straightforward, just change each and every third word for something different instead of what you actually wrote.

And so on ad infinitum.

Yes but: https://en.wikipedia.org/wiki/Consumptive_water_use#:~:text=...

They didn’t need to? They wouldn’t have a particularly bad year.

It's not required, you can re-use the exact same glyph. The letters are effectively the same (both descendants of the Greek 'alpha').

Generally when this happens, it's because they copied the Latin glyph and then made some tweaks to the whole Cyrillic alphabet to make it look more even, etc., and the changes to the 'a' may not even be intentional, but the 'a' got caught up in a batch change or something.

Stopping problem at work. I wonder if results would improve if we gave the agents an “undo” or “revert” tool.

I hope not in the EU.

https://gdpr-info.eu/art-22-gdpr/#:~:text=The%20data%20subje...

The proposed mechanism was hash matching against known CSAM images, so the baptism photos would not trip the filter because they wouldn’t be hash matches.

It’s a bit more like requiring you to submit to a weapons pat down before you go to your locker I think.

I think Sol is pretty much there already

It’s a false idea that a reset “gives you more”. You don’t get any more tokens post reset than before. If you were consuming faster than your allowance before the reset then you probably end up with some free usage, but most people are not likely so far over their pro-rata limit that it’s a huge cost for OpenAI.

What it does incentivise is pushing more usage in case there’s a reset-generating additional demand. Probably sounds good in the boardroom that token consumption is up X% this quarter.

Unlikely to last forever, but for now the goal is to beat Anthropic. Up until the Mythos/Fable debacle, I’d have said Anthropic were leading, but if you’ll forgive the pun, I think they flew to close to the sun, and when Sol came out, they’d shot their “Mythos/Fable” shot and it turned out to be… good, but not revolutionary.

Permafable got announced earlier this week, but definitely the slot machine feel is real.