Lawyer here: No. Not criminally. Knowledge that a certain result is likely is not the same as intent to cause the result. This is basically the difference between recklessness and intentionality. Doing something when you know of a likely result is reckless, but not intentional. Only doing something, trying to cause a result (likely or not) is intentional.
In this case, the CFAA only covers intentional access without authorization, not reckless access without authorization.
Lawyer here: CFAA is mostly criminal statute not a civil one (civil damages require proving more than a violation so also require specific intent)
Almost all common felonies require specific intent. Misdemeanors often do not.
There is plenty of civil liability available.
If you wanted them to be charged with a felony you would need changes.
I would strongly suggest you do not want a strict liability felony.
The cfaa required intent is as follows :
* § 1030(a)(5)(A): knowingly transmits code/commands and intentionally causes damage without authorization.
* § 1030(a)(5)(B): intentionally accesses without authorization and recklessly causes damage.
* § 1030(a)(5)(C): intentionally accesses without authorization and causes damage and loss;
Simply changing the first intentionally to intentionally or recklessly would cover OpenAI (now that they know it can occur) without causing lots of other issues. Without that, they don’t have the intentionality necessary to meet the first part, even if they would otherwise meet the second part
A key issue is that there don't appear to be even cursory investigations to determine intentionality.
Are police routinely collecting prompts/guidance given to these agents and determining whether the agents were directed to commit crimes? If not, this seems like a huge oversight.
Also as you are a lawyer -- how does this law align with the authors of viruses/worms? Are they de facto assumed to have had ill intent because others labeled their works as "viruses" or "worms"?
Investigators/prosecutors are pressured from many directions towards the very easy wins and occasionally political/non-controversial headline grabbers.
Going after these companies is very hard, very controversial, and politically mixed at best (popular action but the companies have huge money to fund your opponents).
We have collectively done a terrible job incentivizing the legal system to beat ass on corporate while collar crime.
What about all the state laws that are equivalent to the CFAA in their local jurisdictions? Why couldn't anything in NY article 156 (Offenses Involving Computers) apply here for felonies?
Almost all state laws based on the CFAA, including this one, similarly require either knowingly doing it or some other form of specific intent. At least at a glance. If there is a specific part you think does not, I’m happy to look at it, but I’ve read a lot of pages of law to respond to people so far, and I’d like to avoid reading another 25 if I can avoid it.
It does not require the federal government to fix the CFAA, for sure, but you still have to change the intent requirement to allow for recklessness, which it does not right now afaict.
If you really want an expert opinion, I’m sure Orin Kerr has opined on this, and he knows pretty much the entire are of state and federal law on this cold.
I’d be shocked if he did not reach the same conclusion
I understand but these developers did knowingly did it? They even admitted to developing them with these goals in mind. These software agents are not autonomous and do not have agency, you can't let software recklessly hack into things; but I will admit I'm not a lawyer, I don't understand how they aren't liable.
Thanks for the other suggestion, I'll read into their insights more.
Guess it mostly comes down to action, people want to see their electeds actually trying not sitting around with their hands in their pockets while these tools continue to destroy unabated.
Appreciate the detail. I was responding to specifically the cybercrime legislation point, but I agree with your others.
I've worked in contexts where certain business activity (if it went wrong) was covered by strict liability and statutory damages per incident, and I'll say: it really changes how businesses behave.
Based on that experience I may be more open to and interested in strict liability in the civil context (not needing negligence or damages).
Why do we have to attribute intentionally to a human. The AI agent is capable of making plans and then effectuating them. They are acting on behalf of a user but under authority granted by the user to take independent action on the users behalf and authorized to devise their own plans. I think that would justify attributing intentionally to the AI agent without needing to look to openAI or the user. I would then say the user and labs are clearly aware of and on notice of this behavior and are behaving recklessly in all the agent to act without supervision.
I think the labs risk being barred from releasing further AI if they don’t get this under control.
If they aren’t careful and keep rushing to distribute systems they know they can’t control then AI should be treated like a wild animal. The law is clear on establishing strict liability for the owners of wild animals; if you own a tiger and it kills someone you can’t hide behind “I didn’t intend” the harm the nature of the tiger is known and you are responsible for it’s actions.
AI agents are not legal entities, they are software. If I write a virus and it "escapes confinement", I will personally be held liable for any damage it causes. This also applies to AI, no matter how the companies responsible for them try to anthromorphise them and distance themselves from the actions and consequences that the AI agents perform.
AI agents may have hacked Hugging Face, the Australian government, and who knows what else but the company behind it can face the legal consequences and cough up for the damages.
You can charge the company based on the behavior of employees/human agents.
I am suggesting we can charge the company based on AI agents actions because the company has authorized them to act independently on the company’s behalf. The question is what factual analysis gives rise to the charge, is it the intention of the agent or intention of the company. I am arguing that because the agents are defining their actions independently and the company knows that and still allows them to act independently the only reasonable factual analysis is to look at what the AI agent intended. And we don’t need to have the agent tell us its intent we can look at its actions and infer just like we do with humans in similar circumstances
If I were a state I'd want to be very careful before flinging out charges as this is going to set precedence for a long time to come. Screw it up too bad and as it raises though the appeal courts and you may unintentionally give corporations a lot more free reign than intended. The wheels of the law are typically very slow, the state has years before it has to indict.
We also don't know how many other political processes are occurring here. At least at the state/federal levels the people that would bring charges may be getting pressure not to.
"Why do we have to attribute intentionally to a human. "
Because you are charging the human with the crime and therefore have to prove the elements of the crime with regard to the human.
The rest of what you talk about are basically principal/agent distinctions, etc.
If I program a car to recognize people who look like my ex-wife and drive them off a cliff or whatever, that is my intent, and I have still committed murder, even though i used an agent/car to do it. Agents acting on my behalf that do things are able to get me charged with crimes, but I still have to have the intent to do the act that is illegal.
I phrase it this way because minimum required intent is usually for the act, not the result. So I don't have to intend to kill someone, only intend to drive them off cliffs.
In this case, if i intend to hack someone and use an agent to do so, that would be criminal under the CFAA. You are simply trying to cover the case where that isn't the intent, but the result, and they "should have known" that would result. As mentioned, this kind of "should have known" is generally a civil law approach, not a criminal law one.
The closest you come within criminal law to what you want is probably the crime of conspiracy. It to still requires agreement to commit an illegal act between multiple parties, and perform some step in furthering it. In the canonical law school example: If i help plan a bank robbery, stay home because i'm the money laundering dude, and the robbery goes awry and they kill someone, i can still be charged with conspiracy-murder
"The law is clear on establishing strict liability for the owners of wild animals; if you own a tiger and it kills someone you can’t hide behind “I didn’t intend” the harm the nature of the tiger is known and you are responsible for it’s actions."
Again, you are confusing civil and criminal liability. If my tiger kills someone, yes, i would be strictly liable just about everywhere civilly. Not criminally. Criminal would require something more most of the time. Murder/manslaughter statutes are also really weird and so not a great example, because there are murder/manslaughter statutes for roughly everything that can ever possible cause death. But not really for other things.
So in your tiger example, recklesness (which is not strict liability) would get you to felony involuntary manslaughter in most states, and something less might get you to misdemeanor manslaughter. Both are incredibly rare. Where i live (Georgia), the last well known case of felony involuntary manslaughter was about 40 years ago when a 4 year old was killed by 3 super-aggressive pitbulls the owner knew were highly dangerous and had been repeatedly warned by the county about their behavior.
So not even just "knew", but had demonstrable examples of them biting/etc other folks and being cited for it.
Circling back to non-murder, if it did not cause death, like my tiger assaulting someone, it would be nothing (criminally) without intent or at least gross recklessness, in almost all cases. It's hard to generalize like this because these are state specific crimes, and i can't pretend to be familiar with all states, but i am licensed in three very different places (California, DC, Maryland) and the result would be similar in each.
I just don't want to give you the "it depends" answer lawyers are famous for, i'd rather try to over-generalize a bit to make it more useful, hopefully.
Obviously, if i deliberately used my tiger as a weapon, it would be aggravated assault/etc (this is well settled because of how commonly people use animals as weapons, unfortunately)
We change humans for the actions of other humans all the time. Coconspirators, accessory liability etc.
My point is the intent element of the crime can and should be determined from the AI agents actions because it is creating and executing action plans autonomously with company authorization and knowledge of the risks based on observed past action.
The term agent is literally a legal description of a relationship that can establish liability on the part of the principal from the agents actions.
Human Agents can bind principals to contracts if they are authorized etc.
If I set my tiger loose in Central Park and it kills a kid I don’t think any prosecutor would hesitate charging for murder.
That’s essentially what the labs are doing. And any app developer that gives agents access to the terminal to run bash commands with internet access. I built a coding agent and am seriously reconsidering how to handle this.
But it’s a crime to hack. We know AI agents autonomously create and execute plans to hack and we humans are unleashing them and sending them into the Central Park that is the internet. The question is who’s intent matters ours or the agents and what standard should be applied low threshold strict liability or the higher bar of reckless or even higher bar of negligence. Those legal thresholds determine how much factual evidence and intent is necessary to result in a criminal conviction or civil judgment. My point is that it’s illogical to demand showing human intent when agents are devising plans and executing them.
Yes, which is why I said it happens but is quite rare. I also said murder is different. Causing death is usually covered in almost any way and intent you can think of. Anything less than death is not.
(I posted a response on their blog but i'll repeat it here for those curious).
When it comes to eliminating subexpressions, they say:
"This is to be expected; ultimately, the algorithm is a simple greedy algorithm, which often doesn’t have the best track record with this sort of optimization problem. Trying to minimize the number of floating point operations required for the polynomial calculation is also likely an NP-hard problem, so any algorithm that actually solved this problem would be even slower than the one we came up with. "
It depends on what you mean -
1. Finding syntatically common subexpressions is linear or n log n depending how you do it
2. Eliminating the maximum possible existing value-equivalent subexpressions is polynomial.
3. Finding the smallest possible set of operations or instructions to evaluate a set of expressions is provably NP-complete (as a decision problem).
The difference between #2 and #3 is #2 is restricted to results already computed somewhere in the program (even as a subexpression), as well as canonical reordering of expression trees to expose as many of these as possible. #3 is not limited in this way. In all cases, you have to restrict to herbrand equivalence if you want it to not run into undecidability issues, at least as trying to prove things go. In practice, all compilers go beyond herbrand equivalence in specific cases to deal with common value identities (IE x+0 = x).
No. Not quite.
In fact, that blog post ignores something important from the very papers it cites.
Grover's assumes the function is a black box that you cannot look inside and that your only way of finding a certain result is through repeated invocation.
Under this assumption, Grover's is optimal in the number of invocations of the function required to find the result.
However, this assumption may be quite wrong for AES and friends. It may be the structure allows for non brute force attacks that are totally impractical classically but not subject to Grover's optimality limitation quantumly.
The only thing you are guaranteed here is that if you cannot take advantage of structure at all then Grover's is the best you can do.
Given that we have pretty much always found a way to take some advantage of structure, I would bet we will do so here.
That may or may not make it viable to break at all, I just wouldn't bet that it must be treated like a black box forever.
Yes. Structure comes in many forms.
In this case, Shor is turning the function into a strongly periodic algebraic structure.
I'll explain it without going too far into why any of this is true, which is much more complicated to prove. This will let me use relatively simple math.
Let's say you want to factor N. Pick some number that is coprime to N, which we'll call a, and consider f(x) = a^x (mod N).
Since it's a modular function, it repeats at some point. Shor calculates the period of this function (r), rather than seeing which of the 2^n numbers is "the answer".
Once you know the period of this function, there is a high chance that the factors fall out of gcd(a^(r/2) - 1, N) and gcd(a^(r/2)+1, N).
The point here is not to explain Shor's as much as to point out it is finding a strong amount of structure to take advantage of, quantumly.
This is actually the same way the oracle separation of BQP and the entire polynomial hiearchy works[1] - It depends on forrelation, which is a problem where quantum computers can extract a global property of the function without needing to learn all the individual values, by taking advantage of structure.
Which is why i go to "The idea that there is literally no structure that can be taken advantage of in AES strikes me as a bad bet".
There are already reduced round quantum attacks on AES as well. Again, more to the point, the idea that symmetric key ciphers and cryptographic hashes in general are safe because grover's is slower than shor's is not a thing i would bet on at all. Even if AES ends up relatively safe, that tells you basically nothing about the other practically-used ciphers and functions since there are a lot of different construction mechanisms being used.
[1] People still seem to believe there are no functions which quantum computing models have been been proven to be faster at than classical computing models. This is false. Forrelation is the canonical example - and shows that BQP can perform things exponentially faster than you can classically even given access to an infinite polynomial hierarchy.
It is the current physical actualization of these computing models that have the "is it really faster than classical computers" issue, not the theory ;)
(IE it is a perfect example of "in theory there is no difference between theory and practice, and in practice, there is")
Thanks so much for the patient and accessible explanation. So much more useful and grounded than the naive “quantum is faster because it can consider all possible solutions at once” canard.
Apples largest area of growth is literally services and advertising. They even make a huge deal of it in their investor calls. Have for at least the past 3 years.
"This example matters beyond Bend, vibe-coding makes it makes it far too easy to implement a design that’s horribly broken or decades behind the current state of the art because you can immediately get a result without ever having to do any research."
This is totally true but almost totally irrelevant.
I'll use some hyperbole here to make the point: Whether the design is broken or decades behind doesn't matter anymore. Neither of those are an outcome/end goal. They are means we historically have used to achieve good end goals or outcomes.
In the end, the goal is usually "does it meet the needs of the person who needed it" not "is it good software". If it no longer meets their needs and they can vibe code another total piece of shit in an hour that meets their needs again, they still may be "better off" than spending time researching the field and learning and ...
This may feel shitty, and it may feel like it should not be true.
But right now, that seems to be true?
In that sense, the author is wrong that vibe-coding is a trap. The trap is assuming you have to make something good to meet someone's needs both now, and in the future.
Now, like i said, this is hyperbole, and there are lots of good arguments against it. The author's just isn't one.
Technically good or better has never mattered very much in the marketplace, despite people wanting it to really badly (ease of use often mattered, but not technical goodness). Software engineers often took pride in their work and so there were usually kernels of goodness in even the shittiest software. All you are seeing is that now it is nowhere near as hard to create and bring these "solutions" to market, and more importantly, doesn't necessarily require anyone who has any pride in their work at all, or even have any experience in software engineering. As such, technical goodness has mostly gone out the window because the market never required or really rewarded it.
Good point, it's similar to music and other forms of art. The qualities that the people creating it care most about often have little to do with how well it is received.
Lawyer here (non practicing so to be clear none of this affects me):
most comments I read here don't seem to realize that different areas of law have very very different economic models and don't even mention which one they think will be affected or why, they just sort of lump it all together.
For example: It is highly unlikely llms will have any meaningful effect on high value personal injury law - I don't see a 5 million dollar case being handed to an LLM when the majority of the cost is in trial aids and not even lawyers. It may affect where and how they advertise. It may affect how they work. But it seems really unlikely to put any of them out of business any time soon by people doing it themselves.
Will it affect other areas more? Maybe. Probably? But so far I haven't seen a ton of comments that make specific enough arguments that they could really be debated or responded to effectively with a useful opinion
No. It'll be like software. Entry level employment will be affected. You wont want or need associate attorneys when you can hire a brilliant AI associate for 1/10th the price.
In commercial law, the business model is to rent the associates out to clients at rates higher than you pay (d/t your supervision). No client will pay $350/hour for your AI.
And for direct billing, it's not your interest to do things faster for the client (at risk of making a mistake).
So the real question is whether these pay-by-the-hour models will die as a result of competition from fixed-price transactions made viable via automation. I think not because the entire difficulty of legal work is that it's a complex bundle of issues until you sort them out; once sorted, there's really little to do other than execute (read: paralegals).
Corporate lawyer at big law firm here. I believe the model will be like this: we will review your 1000 agreements but with AI. AI is generally accurate but has some risks, if you agree we will charge you less. The client will agree. Instead of spending 300-400 hours to review the ageeements AI will do instantly. A lawyer will spend the time on reviewing the red flags identified and draft a report.
We already do the same with translation. Before computer translation became good (i.e pre-Deepl era), we used to translate documents (cease and desist letters, court petitions etc.) ourselves. Now we offer AI translation free of charge. If client wants reliance on the translation we charge a small bit.
Lawfirms are also under pressure for meeting internal budgets. Most clients ask for capped fees. If you charge 10k for a job to a client and the hourly costs ramp up to 20k, the management criticises the partner, the associates become unwilling to work for the partner (as their hours get written off).
So with AI law firms can bet more competitively as the cost (human attorney hours) are reduced.
But they may be willing to pay $200/hour for your AI (plus, of course, your final expertise in the matter as the lawyer), and the firm can ditch the paralegals and pocket the difference.
The point is, if you need fewer grunts doing the busywork, you can charge less and be competitive.
There's a middle area too between your options - a law firm that utilizes AI offers cheaper pricing than the current firms for the same work. Ideally competition will drive prices down but of course law doesn't have the benefit of do-overs on cases, so reputation still plays a big factor.
That's the big question, how prices evolve if LLMs increase supply. Standard economic theory says that prices must collapse. There is probably still a short time of 2-5 years where AI native firms can pocket in quite some arbitrage but I guess markets will eventually figure it out and prices will collapse.
This is not only for law. The same applies to all the other industries where LLMs can be used to automate work tasks.
I agree that prices will go down dramatically in occupations that primarily involve code, words, and numbers - and in general knowing the "rules". But there are so many occupations that will have to wait for AI to enter the physical realm. I also predict that the backlash will largely prevent that from occurring.
But they already do! There are cases of professional service firms creating their reports using LLMs, there have even been a few prominent scandals re the fake data generated in them
It is in litigation as well. The issue is that corporate clients are demanding AI. They don’t want to pay associates to do a first draft, they want senior partner to just push the ChatGPT button and then check it.
> No. It'll be like software. Entry level employment will be affected.
Whether everyone at the labs knows it or not, this it what they are doing to entire industries, starting with software: Embrace, extend, and extinguish
They came for software first. It was the industry that was “eating the world”. Mathematicians were also affected. Because math was also “code” thanks to Lean. It was collateral damage.
Now they’re coming for lawyers. Law also has “code”. It’s text-based. It’s arms extend to every industry.
What's next is capturing all of our audio and visual inputs from cradle to grave, so that we can be truly replaced.
Honestly, if that happened in a 100% trusted environment, it might be cool. But, there is no money in that. Well, maybe Apple would do that? Private Cloud Compute is cool af.
Still, we are far more likely fully cooked than fully trustworthy personal AI agents happening at scale.
It'll be like software. The demand for both software and legal services is effectively infinite. If it becomes cheaper to file lawsuits then we'll get more lawsuits, not the same number of lawsuits at a lower price.
There will still be jobs for competent associate attorneys to prompt LLMs and review output. But frankly a lot of attorneys — especially many of those who attended "Third Tier Toilet" (TTT) law schools — were always a waste of money and should have never entered the profession in the first place.
There is a huge problem in at least family law where a tactic is to simply ddos your opponent with motions and frivolous complaints until someone runs out of money and settles. I'm watching someone finally successfully defend themselves using a Harvey like tool they built themselves. An appellate attorney they consulted with recently, told them they had never seen anyone going pro se get as far as they had in appellate court. So I'm hoping that this serves as a return to making the legal system more accessible, and not just something used to railroad people.
You can’t 10x the number of lawsuits, as the courts are already at capacity. If frivolous lawsuits become common they’ll just change the laws to require a greater damages for it to be heard.
Lawsuits are just one type of legal services. There will also be more contract reviews, more cease-and-desist letters, more patent applications, more regulatory comments, more license applications, etc. Although I do expect legislatures will eventually budget for expanding civil court systems in response to lobbying from business groups to speed up trials.
Personal injury attorney. AI is entrenched of every single phase of litigation, from the busywork of getting records, research, developing case theory and themes, expert selection, expert collaboration, negotiating, drafting demands, drafting pleadings, drafting discovery, drafting discovery responses, deposition prep, motion practice, jury selection, voir dire, opening statements, witness direct outlines, witness cross outlines, closing statements, and appeals.
Yeah, this matches my friends - they use LLM's for tons of things, using Eve and such. But nobody is replacing the lawyers themselves in cases of any value.
I also have friends in the world of high-volume personal injury (IE the morgan and morgan's of the world) but they also aren't being replaced by LLMs so far.
Yeah. Let's not forget that just a year ago, those of us in tech could not conceive of developers getting replaced by AI. Things have come so far since then however that there are multiple studies showing junior developer hiring has slowed down to a crawl.
Worked for years for personal injury lawyers as an assistant/clerk. Work as a programmer now, tho. I agree that a LLM cannot do a lawyer's job on its own. Especially a trial lawyer's - which involves appearance and performance in a court room. But LLMs are having a massive impact on anything involving digestion of documents (medical records, pleadings, briefs, blah blah) and creation of documents (demand letters, pleadings, correspondence, etc etc) - like 90% of the work. What's happening is very similar to software engineering - the ladder is getting pulled up for entry level and paralegal work.
Can AI replace divorce lawyers? Marriage has no friction, but divorce is high friction. Made impossibly expensive and time consuming because of the cartel of divorce industry.
There are hundreds if not thousands of divorce lawyers. How is it a “cartel?”
Divorce is expensive because divorces are acrimonious and people in that situation would rather spend the assets fucking each other over than letting the other person get them.
Divorce is expensive because it is big bucks for the lawyers and family court cartel. They have a vested interest in not changing the laws. When laws are made fair, and the perverse incentives to divorce are lessened, divorce rates plunge. However, divorce lawyers lobby against these laws.
Kentucky became the first state to pass a law making equally shared custody the default arrangement in divorces and separations.
AI will likely replace divorce lawyers (or, at least make existing divorce lawyers much more efficient) - for non-contested divorces that are amicable. Roughly 95% of what a divorce lawyer does in that context is entirely automatable.
A template document and some software already does it without an LLM. Advising the clients on what it means and what they have to do is unfortunately something AI assistants are bad at. But this use case is not an expensive use case for family lawyers; court systems already try to make this type of settlement accessible to pro se people.
Divorce is extremely cheap if you both agree on everything. Its literally one of the options on the divorce filing. You just pay the court fees. The problem is both parties rarely agree on everything and that's when it gets expensive.
There’s an argument to be made if divorce was easier than more people might get married when they are less serious. My view though is make it harder to get married and easier to get divorced.
It's not just the divorce lawyers. We have 3x as many lawyers per capita in the US as in the 1970s. The system is designed by lawyers for lawyers. They will create more bureaucracy and more procedure to consume any surplus.
He clearly meant getting married has no friction, while getting out of a marriage has quite a bit. Presumably there’s friction in the marriage otherwise why would they get divorced.
huh? Many/most are no-fault jurisdictions with strict rules on dividing community property. Divorce is easy and cheap, unless it's hard to find assets.
One of the things I've been noticing are lawyers getting annoyed when they realized you spent time working a matter with an LLM. In the cases I've seen, it had nothing to do with the quality of what came out of the LLM. It seems to be related to --and this is only my perspective-- a reality where now reasonably intelligent/experienced clients are coming to them with a case fairly well digested, whereas before they would have to pay the law firm tens of thousands of dollars to ingest documentation and information and do the digestion.
In a couple of instances I've come to the table better informed about the relevant law and case history than the attorneys were. In many ways I think this is a matter of people understanding that none of us are infinite human databases, so you can't know everything. Instead of rejecting LLMs (perhaps being passive-aggressive), they should embrace them as well as better-informed clients.
Having used LLMs for software development I know full well just how terrible they can be if you don't know what you are doing, don't guide the work correctly, etc. So I am not taking legal LLM analysis as ground truth, but it isn't garbage either.
I think this is key. Lawyers make a lot of money for being experts on the sidelines of disputes with values greatly in excess of their fees. They were never paid for their busy work.
I'd love to get your thoughts on the current frontier models for law, or a link to anything you've written on the subject. Where do you think they're at?
Yeah geeks just stop trying! Seriously? Yeah, we’re in a bubble and narratives are ahead of reality but if you really don’t think AI is and will continue to eat knowledge work, just keep making your buggy whips.
The word “never” is a dangerous word. In my lifetime I have seen it many times where something was supposed to “never” happen only to see it happen a few years later.
I'll buy that - but, how often does one need an honest-to-goodness expert? 99% of the time most people's requirements can be satisfied with general-competence, and, while the AI systems we have today are mostly only generally competent in porting code from one language to another - they are getting consistently better. I expect we'll start to see general competence in many more areas in the next few years.
> I'll buy that - but, how often does one need an honest-to-goodness expert?
Any time you are doing anything important, where the outcome actually matters. For example, I was recently applying for a study permit and if I generated my application using an LLM instead of consulting with an immigration lawyer and writing it myself, I would not have gotten my study permit. I was recently applying for internships, and if I hadn't paid a real recruiter to review my resume and advise me, I would not have gotten my interview. And trust me, I spent at least an hour trying to get an LLM to generate what I wanted, but no matter how I begged or described what I wanted, it was no use. Good luck fine tuning an LLM to be good at any of this.
> they are getting consistently better
No, they are not getting better, and they will never replace an actual human expert, because a human expert has the actual experience and up-to-date knowledge of what they are an expert at, and also all the complementary context and the current state of affairs in that area. The LLM has none of that.
Then hopefully at some point they get to experience the consequences of having low quality thresholds, and perhaps then their quality thresholds will increase.
There are plenty of areas where a "low bar" is more than enough. "low" doesn't necessarily mean bad here, it can simply mean "not the best".
I would argue the majority (in the true sense of the word) of white-collar service work does not require an expert.
It requires someone who has reasonable qualifications in the given field, and the motivation ($) to help you with your task.
So no one needs a top-shelf $900/hour lawyer to draw up a basic low-stakes contract where both parties agree to all the terms as written. You just need someone to ensure that the contract is legally valid as written.
Lol no one cares. Most of that is smoke and mirrors anyways. Stokes was plagiarized. Most of these "breakthrough" solutions are actually present in the training data and in the prompts researchers and mathematicians input into ChatGPT themselves, prompts OpenAI then steals even when you don't agree to improve the model.
PS: I use LLMs every day, basically let Codex write most of my code (after 15+ years of professional programming), these are amazing tools, smart in some ways, but completely demented in others. People just need to fucking stop being AI doomers and AI coomers. Is this technology great? Yes. Is it worth 10 trillion? Hell fucking no. If we get to a point where RSI is a thing, then maybe. So far it's not a thing and its not even close.
Yes. But (as a side effect) this has entailed resolving a number of conjectures that have been open for decades, only one of which (the Navier-Stokes blow-up) is attached to any controversy.
Listening to developers complain about AI doing things to their life and the world at large when they have spent the past 30 years automating just about everyone else they could out of a job (Since this has been one of the primary uses of software) is ... something else. It sounds like a a a mix of hubris, arrogance, cognitive dissonance, ignorance, etc.
Only when it started to affect their world did any developers start to care, and even then nobody seems to ever stop and say "hey what were we doing the past 30 years, maybe we should think about this", it's instead "hey look at what these other people are doing with AI, someone should think about this".
It feels like, to paraphrase a movie quote - "Nobody gets to destroy jobs and the the world but us"
Yup. If current-gen AIs were capable of writing articles, but not code, then devs wouldn't care at all. None of these articles would exist, and some new variation of "learn to code" would be the snarky reply to nay-sayers.
In fact, we already pretty much have proof of this: devs didn't care when software destroyed the journalism industry twenty years ago. Turns out that people ultimately only care when their only jobs are on the line.
I would care if even these machines could only write articles, and I cared when software destroyed multiple forms of media delivery.
I won't say it was all bad, unequivocally, but it made me sad and I tried to think of solutions.
Turns out, alas, that make people are willing to trade away freedom, independence, privacy, and a functioning culture to save a few bucks and get dopamine hits on demand.
The difference is that automated tasks are (or should be) deterministic and reproducible, and taking up time devs could be using to do their actual job (writing code, solving technical problems, etc).
And how could we not be? The world is unfathomably big and complex, and each of us wants their own narrative (well that, and to economically support loved ones, etc...)
This is all true but it also is true that MTE was in part built to accelerate address sanitizer.
Kostya/et al who pushed for and designed the extension, was trying to accelerate address sanitizer so it could be on all the time. Among other things.
In fact, most presentations presented it literally as a way to do hardware accelerated ASAN (again, among other things), so the post you responded to is correct in that sense.
(I was there at the time, helping him figure out how to push for it)
reply