koboll

>I think specifically we want a plugin to be able to indicate if it's safe for it to be triggered without the user's direct consent

It should really be on an endpoint by endpoint basis, since a single plugin can have several different endpoints.

Perhaps the first safety standard OpenAI can implement itself is a warning or blog post or something about how ChatGPT is completely incapable of detecting ChatGPT-written text (there is no reliable method currently; GPTZero is borderline fraud) and often infers that what the user wants to hear is a "Yes, I wrote this", and so it doles out false positives in such situations with alarming frequency.

See: The link titled 'Texas professor fails entire class from graduating- claiming they used ChatGTP (reddit.com)', currently one position above this one on the homepage.

This is the thing that reddit should have done years ago. Great product.

This is so beautiful. Lost it at Marv

>30. Brutal violence against women is depicted in loving detail, often in a story that's ostensibly about violence against women being bad.

Finally, a perfectly succinct explanation for why I can't stand watching The Handmaid's Tale.

Agreed. My heuristic is that GPT-4 is good for compile time tasks but bad for runtime tasks for both cost and speed reasons.

The fundamental problem seems to be that it's still slightly sub-GPT-3.5-quality, and even a long context window can't fix that. It will remember things from many many tokens ago, but it still doesn't reliably produce passable work.

The combination of a GPT-4-quality model and a long context window will unlock a lot of applications that now rely on somewhat lossy window-prying hacks (i.e. summarizing chunks). But any model quality below that won't move the needle much in terms of what useful work is possible, with the exception of fairly simple summarization and text analysis tasks.

They got fined $600k, the telecoms saved billions

Pretty good deal if you ask me

If you read it, a big part of the secret sauce is "actually be thoughtful and informative". Like they drop everything to investigate something and surface useful information and insights as part of the process. Seems like the opposite of spam honestly.

I think you are misinterpreting that line as being the literally true expectation of the author, rather than the author saying what they have to say to minimize repeating of the prompt, which is not really useful output.

Just because that line reads as a person being adamant does not mean that the author really truly believes that the prompt MUST not be repeated and any leak is a security incident. It could just be that the bot has a tendency to talk about itself unless instructed not to, and that's part of the instruction.

>It’s that it didn’t obey what it was told.

I find you basically have to stop thinking of LLMs as software and start thinking of them as unpredictable animals. If you issue a command and expect strict obedience every time, you've already failed. Strict orders are really a tool to persuade certain behavior rather than some sort of reliable guardrail.

Honestly that seems so low priority.

Who cares if the prompt leaks? Really, what harm was done exactly? In what way is the tool at risk?

Two dollars spent over and over would still take a while to equal a software engineer's salary, and if the end result is "produce a fully functional codebase in minutes" the premium might be worth it even if it exceeds that much.

So what's the right way to do a wider-ranging analysis? Chunk into segments, ask about each one, then do a second pass to review all answers together?

It's a lot better than Mastodon, where you simply lose everything.

The author totally glosses over that and scoffs at it being a real problem both in the thread linked and here in the comments, yet real people have experienced the extraordinary pain of watching their accounts evaporate overnight because of some capricious server admin. It's a total nonstarter for ever using Mastodon.

Never really understood what draws people to Cory Doctorow. Over-the-top polemics aren't really good for much except making you angrier about something you're already angry about.

That fourth point is a great one.

One of the most frustrating things in software engineering is eyeballing a decision (i.e. naming a table with a plural) and just gut-level knowing it's the wrong decision, but not quite being able to remember specific pitfalls that instilled that feeling earlier in your career.

That fourth point is the kind of thing that just cuts right through the conversation - "doing it this way does not scale and will fail, here's an example of how". Such frission in finding those and getting consensus around them, because you can feel the future pain you've avoided evaporating.

I'm flabbergasted that OpenAI does not yet offer an API that reliably returns JSON based on some schema you feed it. It's sort of possible to force it to do this but not really to a production-ready degree.

I don't like Reddit's UI for browsing subreddits that are mostly/entirely images, and I don't really like their gallery viewer either. I've always wanted something ultra-simple; just a grid of images and nothing else. So I built it:

https://griid.co/

I don't think anyone else really uses it, and I haven't promoted it at all. But I really enjoy using it.

Skimming has been a valid alternative to traditional reading under certain circumstances since the dawn of the written word; I don't see why this is any more offensive.

Very cool! Demo doesn't seem to work with PDFs though; that would be a useful feature.

Ghostwriter is notably worse than GPT-4, so while it may be true in a sense that "Training a custom model allows us to tailor it to our specific needs and requirements", the reality is they'd be getting better results just using OpenAI right now. Probably true for almost every other use case.

That said, I am patiently waiting and champing at the bit for the day this isn't true anymore. Cool to see the groundwork being laid for it.

Couldn't you argue that "prevent attacks" is a functionality?

Certainly it isn't "squatting" in the typical sense of "prevent someone else from using a valuable name".

Approximately how capable/accurate is OpenAssistant compared with GPT-3/3.5/4?

Edit - Neck and neck with GPT-3.5 according to their own research: https://youtu.be/ddG2fM9i4Kk?t=772

It will absolutely, 100%, be in place as a fallback very soon, and be ubiquitous in that role. Just as AI is now for radiology. That's different from replacing the team of doctors and nurses, though.

Hospitals would love to fire all radiologists and replace them with software.

They've already done it with outsourcing; a large chunk of what used to be done entirely in-house has been contracted out to remote overseas doctors.

>No negativity towards AI here. It’s amazing and it’ll change the future. But we need to be careful on the way.

Yeah, I suspect a lot of fields will have a similar trajectory to how AI has impacted radiology.

It might catch the tumor in 99.9999% of cases, better than any human doctor. But missing a malignant tumor 0.0001% of the time is unacceptable, because it spikes the hospital's malpractice costs. So every single scan still has to be reviewed manually by a doctor first, then by the AI as a fallback.

In theory there's some insurance scheme that could overcome this, but in practice when you have software reviewing millions of scans a day you're opening yourself up to class action lawsuits in a way no competent human doctor would.

>Why do communities exclude manufactured homes? It is the same class-based discrimination that causes local regulators to exclude apartment buildings, and their residents. Manufactured housing bans do the most damage in exurban and rural areas, where they persist far from media scrutiny, certainly compared to recent scrutiny of apartment bans in big cities.

This leaves out a very important piece of the equation. Perhaps the most important piece.

Even in places like New York and California that are rapidly turning YIMBYer, prefab and manufactured housing is being frozen out. The reason is that the construction industry, and (especially in blue states) construction industry unions, are very very powerful and wage total war on legislation and politicians friendly to prefab housing because it completely cuts them out of the process.

The state-level politics are so toxic for this reason that it's unworkable to get stupid bans undone. This dynamic is, frankly, a lot more intractable than the obstacles this article discusses: you can convince regular people that mobile homes are worth it even if they're ugly, but you can't convince construction industry people not to put their own interests first.

Unbelievably frustrating to be a junior or mid level engineer working with a senior dev who does this, watching it happen, maybe even arguing back and being overruled.

I wonder if Goodreads is next under the guillotine.