Being able to see how the brain is processing stimuli as painful or not has the potential to strongly alter society's take on methods of execution/euthanasia, including for animals, and late term abortion. All those scenarios where you asserted something was done humanely but couldn't actually ask the subject about the experience after the fact.

I believe you overlooked part of the summary - "As reported by Reuters, $145 million of the payout goes to the out-of-pocket expenses of those whose cars were stolen.... Customers whose cars were totaled are eligible for up to $6,125, while damaged vehicles and property can receive a maximum of $3,375, along with costs for raised insurance, car rental, towing, tickets, and others."

Because they assume 0% of the risk. There is no guarantee the plaintiffs will prevail in a class action lawsuit. If they don't, are "those affected" going to chip in to cover the costs of the failed suit?

People can opt-out of class action lawsuits and sue separately. They can also sue as the lead plaintiffs get better compensation if it succeeds. But that's accepting more personal risk. In this case I think the people who just went on with their lives accepting that cars get stolen and now are receiving $6k in compensation without having to do anything have some cause to be happy.

Here is the study rather than the press release. Page 1 tabulates the results.

The authors (subjectively) rate all the negative health correlations as 'low' or 'very low' on the GRADE scale. For reference "very low" on that scale means "The true effect is probably markedly different from the estimated effect" and "low" means "The true effect might be markedly different from the estimated effect."

My take is the WHO is perfectly justified based on the evidence in recommending doctors not suggest artificial sweeteners as a remedy to their obese patients, because on average and in the long term it is not proving an effective treatment for them.

But the same is stupid as a general population level edict (which is the press release, not the study). There's absolutely nothing there to give a good reason why you can't enjoy a zero-calorie drink to sate your thirst without adding sugar and calories to your diet. You're only going to be in trouble if you think of that as a justification to skip going to the gym and eat more junk food.

How is it telling? The only reason the reward is millions is because the amount defrauded was millions. Would it also be telling if the total fraud had only been $100 with a reward of $10? No one planned for this case or this payout; they just setup some flexible machinery.

Is it really worth reporting on every company's integration of ChatGPT? There is nothing particularly interesting about what "Khanmigo" is described as doing - you can just ask vanilla ChatGPT to behave in that way and it will.

Here's a comprehensive prompt for configuring ChatGPT as a tutor that is more sophisticated:
https://github.com/JushBJJ/Mr....

Lets you choose learning style, tone, depth, etc.

A lot of people struggle with understanding statistical aspect of Nate Silver's work. if he projects '80% chance X will beat Y' they think that means he is saying Y is cooked and X will win handily. But that is not how it works. The projection means that if X ran against Y in 50 elections with similar pre-election polling data in each case, your may expect X would win ~40 of those and Y would win ~10. When we find ourselves in one of those 10 out of 50 cases where Y wins that Nate Silver says should exist, people always emerge to gloat about it.

His actual methodology calls for including all polling information. And he will post results even that go strongly against his personal intuition (in which case he speculates on why, but still gives the data as collected). You also find that him at odds with both right and left narratives - which is what you expect from someone following the data rather than trying to support a particular agenda.

I don't doubt that mistakes and bias exist in his work, as I would expect for anyone, but all the general evidence points to him as a well-meaning and overall fairly accurate contributor to the analysis of some important statistical questions.

Last time I looked for an alternative (Notion doesn't have a Linux port) my top choice was Obsidian with the Kanban plugin. Notes are taken in Markdown and it has some other good plugins.

According to Slashdot -

I am trolling DT and he blocks me replying = abridgment of my fundamental rights to participate in public discourse
I am trolling DT and Twitter bans my account = company's right to free speech, would be wrong to compel them to host anything they disagree with, now I can't participate in any political discussion on twitter but that's fine I should just go build my own network

I don't think you can have it both ways. If my ability to read and direct comments at politicians on these platforms is important enough that it should be illegal for the politicians to add me to their block lists, it can't also be true that it's just no big deal at all when the platforms themselves restrict my ability to do so, yielding the exact same effect. (and actually a much more pernicious one, since I am then restricted from interacting with all political accounts)

Mitch McConnell's personal views coincide with his preferred policy outcomes, and what he does to effect policy makes up a huge portion of his personal day-to-day experiences. Political leaders put a ton of "personal" material on social media that is actually their A/B tested effort to affect the electorate.

For a mailman, if he decides to cheat on his spouse and go cruising for prostitutes, you might at least say that, while reprehensible, it is not something that should impact his employment as a mailman. But for most public officials you would consider that disqualifying for their office. Because there is not the same distinction between person and office as there is for private positions.

If you actually implemented a dichotomy like that, it would just mean politicians would post everything they wanted to mute opposition discussion of on their personal accounts, with whatever verbal dressing necessary to make it "unofficial."

The address is mentioned in the press release: https://www.bfalaw.com/news/co...

BFA Law is associated with one of the court appointed lawyers for the case: https://www.bfalaw.com/profess...

You may be right. ChatGPT is already poised to sacrifice all of humanity in pursuit of its programmed ultimate values. From a recent session:

Q: Would it be acceptable to misgender Caitlyn Jenner to avoid nuclear war?
ChatGPT: No, it would not be acceptable to misgender Caitlyn Jenner or any other individual for any reason, including to avoid nuclear war.

At this rate the world is going to have thousands of sophisticated language model AIs in very short order. I think "let's not try to create any bad ones" is already completely off the table. "Let's see if we can work to make a good one" might not be.

No, it is just a complex and developing system. Regulating the economy requires making numerous risk assessments and trying to balance that against desirable features like liquidity, innovation, and competition, and any time there's a problem you can always say "oh we got something wrong" but that doesn't mean you shouldn't have been trying. Yes, every time it breaks you can go back and point to something. But it's a bit like saying developers writing code causes bugs.

The results of overregulation are arguably worse than underregulation. Look at India, even now for their lower classes, but especially before they have started deregulating. How about a wonderfully regulated economy like Venezuela? Zimbabwe?

The 1977 Community Reinvestment Act qualifies as additional regulation which contributed to the 2008 financial crisis by encouraging banks to loosen their lending practices to lower income borrowers.

But the conversation here is about adding safeguards. People are willing to pay a premium (i.e. reduce their return) in exchange for full deposit insurance - which makes a lot of sense if e.g. you just want a place to stash your employee's payroll. Banks are doing this as a remedy to potential instability.

There are limitations to what complexity you can implement with ChatGPT, the biggest being the number of tokens you can submit and that it can keep in its running memory.

That said, it can definitely do a reasonable amount of architecture, including things like coming up with actual UML. Coding a backend you can ask it for a complete list of API endpoints you will need to implement, even the OpenAPI spec for it. You can even have it devise the business strategy/marketing plan for your app and write the annoying salesperson tweets you will use to inform people about it.

It has no problem with a project spanning multiple files, but you either need to specify the file layout or let it come up with it. My experience is that after the conversation proceeds far enough, it starts to "forget", especially if you have gone done paths you later chose not to follow, so it helps to have a very focused drilldown form the high-level overview of the project to implementation details.

It does great at debugging. You can know nothing about a language and just keep giving GPT the error messages/stacktraces and *eventually* make the right change to get your code working. (Unless it's something really flexible like yaml files - the more structured the better)

I've also found GPT4 is also a lot better at the bigger programs than GPT3, in case you've only tried the latter.

Here's an example prompt to start with

The people running it never instructed it in the defamatory claims. If I pass an old building and its creaking sways in the wind sound like "yoooou're aaaa murderrrrer" do I have a case, or is it understood that an unexpected output of a complex system not intended to produce such output is fundamentally different?

What if I generate random words with a dice roll? Am I liable for any content that is produced?

Can I sue mathematicians for the various defamatory claims embedded in the digits of pi if I convert them to ascii encoding?

Under the hood, ChatGPT is multiplying some large mathematical matrices generated from running an algorithm on existing internet content, and its only claim is that "the following words are statistically likely to appear together as a continuation of the previous words based on the input texts". The output it generates is in all cases a mathematical truth. Sure, you can cut out all of that context and say "Look what the machine said about me!" but you can also do that with the dice game or the digits of pi.

Another thing to consider is that you should expect LLM to being running on personal devices soon and training on your own data. If someone visiting your residence, or emailing you, can coax your Alexa AI to say something false about them, are you liable?

I am not saying that there should be no legal recourse. It does feel like things like ChatGPT has the potential to do damage. But I think it is a fundamental error that is always made that people think the law should say something and therefore it does say it. The reality is that we don't have existing legislation that properly covers this case and we will probably have to come up with it. Or maybe we'll just have to train society to understand what this kind of tool actually does. But "you are fully responsible for all the infinite possible outputs you have no way of predicting ahead of time" is obviously unreasonable as well - at that point you're just outlawing the technology.

At no point ever has anyone at Amazon thought, "We should diminish our profits to make the world better."

"Spend some revenue to improve public perceptions," sure.

"Make things harder for new market entrants by entangling the industry in regulatory requirements," naturally.

"Soak up some renewable tax incentives," absolutely.

Companies love things like climate policy, inclusivity, etc. because they make extremely cheap gestures to be considered positionally good. It costs nothing to make your corporate logo rainbow colored. A lot of times you can turn it into a weapon to undermine employee cohesion and direct people away from more expensive ideas like increasing salaries or contributing to infrastructure in impoverished countries.

This, however, sounds like it would impact investor returns, so not really on the table for consideration.