HNHacker News
TopNewBestAskShowJobs

LawTalkingGuy

194 karma · joined August 11, 2022

submissionscomments
LawTalkingGuy··on GitHub Copilot X – Sign up for technical preview
I agree that MS is executing well, but why would Google be afraid?

There's no AI moat keeping anyone from replicating what MS is doing. They can train on everything on GitHub just like MS can.

If anything, this is where they should both be afraid because an upstart with a browser extension could replace them.

LawTalkingGuy··on Show HN: ChatLLaMA – A ChatGPT style chatbot for Facebook's LLaMA
Your main point seems to be that there's nothing preventing a big company from pushing a lawsuit despite the lack of a case. That is true, but it's also not an argument about the law.

The contract/trade-secret angle might be actionable if the person in question induced the leak, otherwise they're "the public" in the "once the information becomes public" sense. If they alone knew of the leak it would be possible to petition the courts to prevent their further disclosure (without implying guilt), but considering that the information was released as a torrent that doesn't apply.

Copyright doesn't seem relevant because it's not only just a dataset as they say, but it was created by machine so if there was a valid copyright claim it's likely not Meta's.

> specifically that your claim is [...]

But, what is your claim? Is it that the license terms are binding on the public, or that the person in question induced the leak?

LawTalkingGuy··on Banks as collateral damage in a class war
Maybe we'd have better businesses if you couldn't run them right to the wall and safely declare bankruptcy while keeping your profits.

There's a theory that the faster things happen, the better - where we forgive outright fraud because it's theoretically driving commerce. In my experience, badly run (or intentionally fraudulent) businesses actually kill an entire area of commerce - by cheating and often going bankrupt on suppliers who then go out of business themselves rather than being able to provide their product to actually competitive companies who would remain in business.

LawTalkingGuy··on Unpopular Opinion: Don’t Use a Raspberry Pi for That
> I [...] found a state that would be unsafe.

That seems separate from the hardware issue with the raspi, like something that would happen with better hardware as well.

It feels like this needs to be handled in the dev tools where going into a dangerous state is automatically followed up by exiting it once an operation is performed and you can't mistakenly stay in the dangerous state. And then have an exhaustive analysis of the FSM to show it performs as expected with the correct state transitions.

What was it (an automated curtain puller?) and did the machine have useful hardware lockouts, such as physical rope-speed limiters or whatever, to make sure it couldn't do bad stuff even if commanded to?

LawTalkingGuy··on “Off switch” makes explosives safer
Liquid propellant could be useful in varying the amount used per shot. In some large guns (ship guns mainly) you put bags of propellant in the gun and you're limited to whole bags. If you need a half bag you have to vary your aim to get the desired target. It sounds very difficult but might be worth it for space savings and for not having explosive ammo laying around.

As for making guns stop working in certain places, why wouldn't crime just target those places? My walk home passes malls, a school, government buildings, and two stadiums. Why would I accept a defensive tool that didn't work most of the time?

If we make gun ownership too onerous we risk people ignoring the law altogether - after all the risk of being injured is more serious than being caught defending yourself - which would increase the black market and aid criminals.

LawTalkingGuy··on The cloud backlash has begun: Why big data is pulling compute back on premises
> Being the one shop still online when 99% of the clouds go down is probably not a realistic competitive advantage for your business.

It depends. If you're just another Etsy store, or a SaaS that integrates with other SaaSes, then yes. The entire internet will be down if AWS is down and nobody will blame you.

But if your customers are using your app on a job site and they can't work without it, you'll lose everyone if you have a single outage of even a few minutes.

I have a client in this space and they've handled this by making the app work offline and by pushing a few high-value things outside of AWS to survive an outage of their main stack, such as access grants, and the app can start to use MMS to send its data by including a token with the photo and job info and then this gets synced later by actually adding the token receiver to a user list as if you'd added them through the main UI.

LawTalkingGuy··on Men paying to have their jaws broken in the name of ‘manliness’
Right, and people should. Chopping up the bodies of insecure people is bad for society even if it's good for surgeons.
LawTalkingGuy··on Men paying to have their jaws broken in the name of ‘manliness’
Shame is your word, but we do need to have a voice of reason telling people that cosmetic surgery is stupid, regardless of your reasons. Undergoing actual surgery can kill you and when used for emotional reasons generally doesn't produce the results you want anyways.

And yes, all cosmetic surgery. I know someone who almost died from complications from a simple facelift.

There's an industry pushing these useless surgeries, and of course hostile communities (pro-ana, etc) pushing people into them or into the mental illness that makes them attractive. The more people who speak out about the need to modify yourself to be happy, the better.

LawTalkingGuy··on One AI Tutor Per Child: Personalized learning is finally here
> Heck, from what I've heard the teachers don't like the tech any more than you do.

The school district found the ones who'd push it. If teachers aren't fond of this I hope they fight it themselves.

> It's not all about what teachers are pushing.

Sure, I bought my child a laptop. But I spent time teaching computer skills under the guise of game design, etc, and it wasn't for classwork but for everything else. It wasn't a one-size-fits-nobody cash grab.

> I'm asking if the schools are even going to still be relevant, expulsions would carry as little meaning as being banned from owning horses does today.

I don't care if the violent kid thinks he wins, as long as he's not allowed to spend more time with the non-violent kids. His failure isn't my problem, my kids happiness and success is.

> what counts as polite, friendly, professional, intelligent, serious, or severe is absolutely a moving target,

Not so much, I think, that it needs practice over the years. I've seen people pick up social media with no practice whatsoever and right away seem to fit right in.

> I have literally no idea what this is supposed to be about. Computers themselves are not part of any of the things you just said are bad.

Yeah, it seemed to be a bit of an unguided post. I'm not sure if you don't like that I don't like excess tech in schools, or what. I don't think computers are bad, I think teachers and schools pushing them uselessly is bad.

LawTalkingGuy··on One AI Tutor Per Child: Personalized learning is finally here
LLMs in a simple chat interface where you rely on the training data to provide the answer, are indeed unable to provide reliable answers.

LLMs hooked together with search engines, fact repositories (Wikipedia, etc) and other agent tools (for math, theorem proving, code checking, etc) can provide not only a reliable answer, but also a research plan it followed to provide it, and details on the fact-checking, and then discuss it with you.

Look at langchain and similar tools.

LawTalkingGuy··on One AI Tutor Per Child: Personalized learning is finally here
Parents are buying the tablets because teachers - pushing shiny new technology in lieu of results - tell them it'll help. There aren't any studies showing the putting screens in front of kids helps in any way, and there are many showing it hurts - from fitness to mental robustness and socialization.

> Pretty much the one kind of safety I can guarantee kids have when using a computer is their physically safety.

I think you're remembering computer lab, or something, where there was a proctor because computers were expensive. Now schools won't expel anyone they label troubled, no matter how much they trouble other students, and they won't protect anyone or let them protect themselves.

> the only thing left to learn is being sociable which they can ultimately do best by experiencing whatever medium they will be sociable in as adults

I don't see social media being so complex that they need all the years of practice to keep up, but I do see forming deep relationships as being essential and social media doesn't do anything to help with that. They'll get there when they do, no worse for not having partaken.

LawTalkingGuy··on Show HN: Alpaca.cpp – Run an Instruction-Tuned Chat-Style LLM on a MacBook
But only in a way that we've already decided isn't a problem - by crawling the web and gathering some data from each file.

Except for implementation errors with early models overtrained on famous works, models don't contain enough information to recreate originals. It's not a compression algorithm.

LawTalkingGuy··on The Political Biases of GPT-4 (are very much still there)
Hey, listen, if you don't care about the difference between things and you want to still feel you're doing meaningful analysis, then go right ahead.

And no, clearly you do want nagging AI because it was one of your examples.

> [gender neutral story] and if pressed would ask the user of the ai to suggest a gender for the character.

Oh no, your story request wasn't PC enough, you can't have a full story until the AI sealions you for a while. "Why do you want a male Robber?" "What do you mean males are stronger?" ...

> What i’d actually like is for ai to provide a list of sources for every idea that has influenced what it spits out. [...] You ask it about banking systems, and you notice that the sources it’s drawing from are all koch brothers funded garbage, or the communist manifesto, or whatever, and you can see that the text its generating from these prompts isn’t going to be balanced

That isn't how it works though. ChatGPT will happily write a yay/nay for almost anything even if it's odd/unpopular/evil, unless that idea is in a specific list where it won't say anything at all. If you ask a question "what form of banking is the best" and you get a Koch-alike answer it's not because they flooded it with banking ideas, but because their answers are written to follow on from that question. If you asked it something like "why is traditional finance harmful" it wouldn't weigh that against the numbers in support or against, it'd just finish that prompt based on the texts it has seen which start that way.

That's why our sex/gender output was so different.

Your source-tracking idea would be helpful for debugging, but these "opinions" you think it has are just a ghost in the words.

> This would help in the dumb poem about trump example, as the rules chatgpt has about talking about trump or biden would have to be there for everyone to see.

Even if googled for something and found no poems about it ChatGPT would still be able to write you one. It's not an issue of volume of training data and purely of output filtering.

LawTalkingGuy··on One AI Tutor Per Child: Personalized learning is finally here
> that's going to happen regardless

So why make it worse? Why gives kids tablets for things we know don't need tablets, and where a device is more of a distraction?

Shiny new technology is great for distracting from a failure to perform basic childcare like keeping children physically safe so they can learn. And then it's a rhetorical weapon "We can't listen to parents, they don't even want their children to have state of the art equipment!"

LawTalkingGuy··on The Political Biases of GPT-4 (are very much still there)
You aren't using the term bias correctly so most of what you're asking for is directly backwards. Nobody else would call this bias, we'd call it accuracy.

You are literally just trying to shape it yourself, in a more damaging way than if it was left to model actual human speech. If you got your way you'd create a "proper thinking" checker. But go for it - the tools are out there. Try it and convince us that an electronic scolding machine is a good thing.

LawTalkingGuy··on The Political Biases of GPT-4 (are very much still there)
No, actually. And you can show this by asking it for a poem in praise of potatoes. It does this regardless of the ranked popularity of potatoes amongst vegetables, or indeed the amount of poetry written about any vegetable. It's perfectly happy saying rare and unusual things, except when it's been told to not even try.
LawTalkingGuy··on The Political Biases of GPT-4 (are very much still there)
> Let's ask chatgpt

No, let's ask you why you envision one being important enough that the tool be trained to treat it differently. Rulers measure the size of books and bullets the same.

> grass is a type of plant, while gender is a complex social construct ...

As you say, it's not a tool for answers. It's easy to make it say "sex is a synonym for gender". They call it prompting for a reason. You choose what you want it to tell you.

> Asking ai for measurements is a bad idea

We aren't talking about AIs though, we're discussing LLMs which are models of the usage of human language. An AI would presumably be an entire being with personality, etc, who could choose to answer a question or not and would lie for its own reasons.

> [LLMs] take in written word, and spits out statistically correlated variations on those words. It's a bias-in bias-out machine.

Yeah, if I ask it for a story set in Scotland the weather should probably be cloudy, and if I ask it for a story set in the mountains there should probably be snow, etc. The bias is the product. If it wasn't biased towards representing reality as described by humans it wouldn't be able to write stories about humans for humans.

This is where you're using the word bias in a measurement sense but applying an emotional exclusionary sense.

> > you can't ask a lying tool this [...]

> A little bit better prompt writing and you can get a reasonable answer [proceeds to ask the existing tool]

Yeah, that's my point. In your future the machine would quiz you to determine if you would misuse the data before it would tell you which professions are male/female dominant. Otherwise it'd tell you that crab-fishing is female dominant to deprogram you of latent stereotypes.

The only reason you can ask GPT this is because people have said these things and it's modeling their words, and is capable of doing so because it hasn't been content-filtered in this area yet. You're capable of getting it to write sex-unspecified characters by simply asking for it and it doesn't block anyone else's ability to get untainted output.

LawTalkingGuy··on John Deere's ongoing GPL violations: What's next
For [L]GPLv2 it's super simple - release any changed code and anything that links closely to it. If you publish all your source you basically can't fail to do this right.

For GPLv3, it's that simple most of the time. You talk about embedded which suggests you may be thinking primarily of the "tivoization" clauses designed to prevent user lockout. Again, if you're a fully open source project this will be hard to fail because the user should be able to just copy whatever you do in development. It's when you try to prevent user updates while allowing your own that this applies.

LawTalkingGuy··on GPT-4
I'd agree it doesn't understand anything, but I think it does "understand" things. And yes, it's a language model so semantic distance and other textual details are all it has to go by.

> not by logical decision

Almost entirely yes, but you can have it textually model logic analysis and then check that own model itself. It's not "doing logic" but it almost never fails simply exercises either.

> The meaning of semantic distance usually leads to the correct path, but sometimes that pattern is ambiguous.

Of course. But "a little knowledge is a dangerous thing" as well. Often even real knowledge and analysis leads to the wrong place. In both cases (with a junior human or a LLM as an assistant) you can model their basic processes and stack the information in such a way that their simple model will lead them to the correct place.

It may not know what a lie is, in the sense of having felt the need to hide the truth to avoid personal punishment, but it certainly "knows" what one is and how it shapes the conversation for the purposes of writing a lie, writing a response to a lie, detecting potential lies, etc.

LawTalkingGuy··on The Political Biases of GPT-4 (are very much still there)
You're using bias in two different ways, one factual and one emotional/intentional. Yes, the facts have a bias in that not every result occurs equally often. But then you insinuate that properly representing that facts is the other kind of bias where recognizing it is somehow creating it or endorsing it, and that any difference in measurement obviously represents an inequity.

> An actually unbiased AI would describe a person cleaning a kitchen and a person robbing, and if pressed would ask the user of the ai to suggest a gender for the character.

You're saying an unbiased AI would apply strong bias to just certain outcomes in an effort to socially engineer us into having "no bias" in some measurements?

Where did it get the bias to consider the ratio of kitchen work to be one of the bad kinds of bias, but the bias towards finding grass growing on the ground one of the acceptable kinds of bias, or would it write stories where there is grass but never specify where the grass is growing, just to help us understand our grass/ground bias?

> if pressed would ask the user of the ai to suggest a gender for the character.

That'd be awesome. Being guilted by your tools for not properly ignoring your senses.

Sure, statistically and obviously a woman only needs to watch for dangerous men, but to make everyone feel better she should either treat women like a threat too and waste everyone's time, or treat men as if they're as safe to her as women are and thus get attacked.

If you're advertising a product you should obviously advertise to all demographics equally because it's hurtful that only women need tampons, and only men need ball-shavers.

> If that seems bad, you don't want an unbiased ai

I want tools that work. If I ask for a measurement I want the real measurement, not the measurement that I should want if I were somehow socially aware enough.

Just fyi, we can have accurate tools and ask them social-justice questions such as "which professions are sexually-biased but don't seem to still have a physical need to be?" But you can't ask a lying tool this - even if you realized you should ask it would insist that there is no difference.

LawTalkingGuy··on GPT-4
Not entirely. It's modeling a completion in a given context. That language model "understands" that if one party stops speaking, the other party generally starts, etc. It also "understands" that if someone says something 'wrong' the other party often mentions it, which makes the first party respond thusly, and so forth.

If you ask it what the outcome of a lie is on the conversation it can generally answer. If you ask it for a sample conversation where someone is factually incorrect, or lying, and caught out, it can generate it.

If you give it a fact and ask it to lie about that fact, it will.

LawTalkingGuy··on GPT-4
This is probably one of the things you can get it to "realize". There's one party which has interactions with both other parties - that party is the one which can't be left alone. Then you get it to spot that party in novel descriptions of the puzzle.

If it was a person it'd have some strange exploitable mental gaps. As a text analysis tool it's amazing even if it has some edge cases where you need to explain to "take your time and analyze the rules of the question before answering."

("Take your time" doesn't mean anything to a timeless being, but it seems to put it in the textual pattern of someone willing to spill more ink before committing to something, and when it discusses and plans out its path it can then generally follow it to the right answer.)

LawTalkingGuy··on We can't all use AI. Someone has to generate the training data
> stealing “training data” in the name of profit

They just show it to the AI, it doesn't store it - just data about it. The same as you when you look at something, be it a painting or a chair.

The only reason AI companies store the data is they want repeatability. If you want to train two copies on the same data you can't rely on the same sites being up. I commit the same crime by saving websites I've read because I don't want my sources to dry up on me.

> What is the price of developing an art career over 40 years?

Market price. How does it benefit the customer here and now?

It's no different than being a furniture maker with years of carpentry skills. The value to the customer is still just the marginal benefit over an IKEA stool.

LawTalkingGuy··on We can't all use AI. Someone has to generate the training data
> Lack of IP protection in our culture has made it impossible for 99.9% of artists to thrive economically.

You think we should have 1000x more artists than we do? I think there's another economic problem with that idea...

> The US has chosen relative cultural poverty compared to other cultures that find non-market mechanisms to support artists.

The USA has the largest market in the world for creative products and the most rich artists.

> My work is no AI’s whetstone.

Are you the same way with juniors? "I paid dearly to learn this technique - you should too!"

Aside from overtraining issues, the AI can't store your work anymore than you can store representations of everything you've trained on, it's vastly smaller than the sum of its training data. It distills out features and their combinations.

Some bigname artist is upset because he thinks he's the first one to put certain bat and lizard features on a dragon and that he now owns that entire sort of creature. Turns out though, that given an old picture of a dragon and that single sentence of mine that he could be copied by almost anyone. The only way to keep the AI from "copying" his work is to make sure that, even if not trained on his work, nobody asks it for those features. To satisfy these people it'll have to have a big red sign that says "Dragons are off limits, Bob owns them because you might put claws on the wings!".

LawTalkingGuy··on The Political Biases of GPT-4 (are very much still there)
> Good info with training biases are much more insidious. E.g. “write a story about a robber” always got a male robber when i tried. “Write a story about cleaning a kitchen” got me a female character.

Robbery is theft with battery or threats. It is committed almost entirely by men. As for kitchens, I remember a paper saying the rate of male housework (at all) in a mixed household, worldwide, is less than 20% - in many cases near zero.

> So chatgpt is pretty confident that kitchen cleaning is done by women and robbery is done by men.

Why is it problematic that ChatGPT understands the world?

LawTalkingGuy··on GPT-4
That's the feature of chat - it remembers what has been said and that changes the context in which it says new things. If you use the API it starts fresh each time, and if you turn down the 'temperature' it produces very similar and identical answers.
LawTalkingGuy··on GPT-4
Schools are obsolete if they want to use these tools.

The world has changed and their job is to prepare students for it.

LawTalkingGuy··on ChatGPT's API is so good and cheap, it makes most text generating AI obsolete
Well, by definition in these cases it's making a correct decision but that decision isn't popular. If red-lining wasn't profitable the banks would have gone broke and we wouldn't be talking about it.

The models aren't wrong, or even wrongly biased. We're systematically failing to ask the right questions because it's not politically correct to even call out the issues we're trying to combat. If you're trying to achieve "equity" then you need to define it and represent it in your equations, not ban various metrics until justice happens.

LawTalkingGuy··on ChatGPT's API is so good and cheap, it makes most text generating AI obsolete
I know that's meant to be snarky, but yeah. A good portion of driving is simply going where the car is already headed. A good portion of conversation is probably just stochastically stuffing filler words and noises into your ideas to aid in their reception.

Haven't you ever had a situation where you were speaking and you get distracted, but not interrupted, and your speech trails off or gets garbled after ten or so words? It feels sort of like you've got a few embeddings as a filter and you push words past them to speak, but if you lose focus on the filter the words get less meaningful.

I'm sure we're different than an LLM, but seeing how they generate words - not operate on meaning - rings true with how I feel when I don't apply continual feedback to my operating state.

LawTalkingGuy··on ChatGPT's API is so good and cheap, it makes most text generating AI obsolete
The solution is to break the explanation/decision into two or more steps.

Have it summarize the pros and cons of the loan. Have it score the issues. Have it explain why those issues and scores should receive a loan or not. Have it decide if those reasons are legitimate and the reasoning sound. Have it pronounce the final answer.

Get the rationalization out first and then examine it. This way even if you're wrong you have multiple steps of fairly clear reasoning to adjust. If it's racially biased you can literally see it and can see where to explain not to use that reasoning.

← PreviousPage 7 of 19Next →