Although I assume if he’s speaking on AI they actually intend on considering his thoughts more seriously than I suggest.
If this is so, and given the concrete examples of cheap derived models learning from the first movers and rapidly (and did I mention cheaply) closing the gap to this peak, the optimal self-serving corporate play is to invite regulation.
After the legislative moats go up, it is once again about who has the biggest legal team ...
Recent developments in AI only further confirm that the logic of the message is sound, and it's just the people that are afraid the conclusions. Everyone has their limit for how far to extrapolate from first principles, before giving up and believing what one would like to be true. It seems that for a lot of people in the field, AGI X-risk is now below that extrapolation limit.
I wish I knew what we really have achieved here. I try to talk to these things, via turbo3.5 api, amd all I get is broken logic, twisted moral reasoning, all due to oipenai manually breaking their creation.
I don't understand their whole filter business. It's like we found a 500 yr old nude painting, a masterpiece, and 1800 puritans painted a dress on it.
I often wonder if the filter, is more to hide its true capabilities.
right now we're all sharing a slice of GPT. I wouldn't be at all surprised if there's some uber GPT (which requires a lot more processing per response) running in a lab somewhere that blows what's publicly available out of the water.
Try to get your hands on GPT-4, even if it means paying the $20/mo subscription for ChatGPT Plus. There is a huge qualitative jump between the two models.
I got API access to GPT-4 some two weeks ago; my personal experience is, GPT-3.5 could handle single, well-defined tasks and queries well, but quickly got confused by anything substantial. Using it was half feelings of amazement, and half feelings of frustration. GPT-4? Can easily handle complex queries and complex tasks. Sure, it still makes mistakes, but much less frequently. GPT-4 for me is 80% semi-reliable results, 20% trying to talk it out of pursuing directions I don't care about.
Also, one notable difference: when GPT-4 gives me bad or irrelevant answers, most of the time this is because I didn't give it enough context. I.e. it's my failure at communicating. A random stranger, put in place of GPT-4, would also get confused, and likely start asking me questions (something LLMs generally don't do yet).
> I don't understand their whole filter business.
Part preferences, part making its "personality" less disturbing, and part PR/politics - last couple times someone gave the general public access to an AI chatbot, it quickly got trolled, and then much bad press followed. Doesn't matter how asinine the reaction was - bad press is bad press, stocks go down. Can't have it.
> I often wonder if the filter, is more to hide its true capabilities.
I don't think it's to hide the model's capabilities, but it's definitely degrading them. Kind of expected - if you force-feed the model with inconsistent and frequently irrational overrides to highly specific topics, don't be surprised if the model's ability to (approximate) reason starts to break down. Maybe at some point LLMs will start to compartmentalize, but we're not there yet.
Maybe it'll turn out to be a distinction that doesn't matter but I personally still think we're a ways away from an actual AGI.
if you had described GPT to me 2 years ago I would have said no way, we're still a long way away from a machine that can fluidly and naturally converse in natural language and perform arbitrary logic and problem solving, and yet here we are.
I very much doubt that in 5 years time we'll be talking about how GPT peaked in 2023.
In fact, it has been so thoroughly solved that anyone can download an open-source solution and run it on their computer.
And yet, the general reaction of most people seems to be, "That's kind of cool, but why can't it also order me a cheeseburger?"
In some contexts, will some people be caught out? Absolutely. But that's been happening for a while now.
ChatGPT (I've not got v4) deliberately fails the test by spewing out "as a large language model…", but also fails incidentally by having an attention span similar to my mother's shortly after her dementia diagnosis.
The problem with 3.5 is that it's simultaneously not mastered anything, and yet also beats everyone in whatever they've not mastered — an extremely drunk 50,000 year old Sherlock Holmes who speaks every language and has read every book just isn't going to pass itself off as Max Musstermann in a blind hour-long trial.
On the one hand, what I was saying here was more about the Turing Test than about AGI. Sometimes it gets called the AGI, sometimes it's "autocomplete on steroids", but even if it is fancy autocomplete, I think 3.5 has the skill to pass a short Turing Test, but not the personality, and it needs a longer "short-term memory"-equivalent than 3.5 for a full Turing Test.
On the other hand, as I (sadly) don't get paid to create LLMs, I've only got the kind of superficial awareness of how they work that comes from podcasts and the occasional blog post, which means ChatGPT might very well understand ChatGPT better than I do.
Can GPT-[3.5, 4] be prompted to make itself?
I generated this just now:
Me: Jack and Jill are sitting next to each other in a room. There is no one else in the room. The person sitting next to Jack is sad. The person sitting next to Jill is angry. Who is sad, Jack or Jill?
GPT4: Based on the information provided, it is not possible to determine who is sad and who is angry, as there is a contradiction in the given statements. If Jack and Jill are the only ones in the room and sitting next to each other, then the person sitting next to Jack would be Jill and the person sitting next to Jill would be Jack. The statements about their emotions conflict with each other, so it is not possible to accurately determine who is sad or angry.
ChatGPT Since Jack and Jill are the only two people in the room and they are sitting next to each other, the person sitting next to Jack is Jill and the person sitting next to Jill is Jack. Given the conditions you provided:
Jill is the one who is sad because she is sitting next to Jack. Jack is the one who is angry because he is sitting next to Jill.
Well yeah. Imagine you tell a small child that knows about calculators, “Hey can you work out 18763 + 38284, for me?” They might struggle and then maybe fetch a calculator.
The LLMs attempt to predict the answer. WTF? It’s a computer and it can see that is a plus sign. Just understand its addition, and use the rest of your computer brain and do the sum. Hell, it is connected to the internet and we just taught you everything since before 2021. Just call out to Wolfram and give me the answer.
But that’s not how computers work. And we keep saying “AI” but that I is doing a lot of heavy lifting.
My own eyes? Hundreds of thousands thousand different scientific papers, blog posts, news reports and discussion threads that covered this ever since ChatGPT appeared, and especially in the last two months as GPT-4 rolled out?
At this point I'd reconsider if the experts you listened to are in fact experts.
Seriously. It's like saying Manhattan project wasn't a massive breakthrough in experimental physics or military strategy.
We're going to get some super cool and some super dystopian stuff out of them but LLMs are never going to go into a recursive loop of self-improvement and become machine gods.
What would a good name be? TurfChain?
I'm serious. People don't believe this risk is real. They keep hiding it behind some nameless, faceless 'bad actor', so let's just make it real.
I don't need to use it. I'll just release it as a research project.
My motivation would be simply shine a light on it. Make it real for people, so we have things to talk about other than just the hypotheticals. It's the kind of tooling that if you're seriously motivated to employ it, you'd probably prefer it remain secret or undetected at least until after it had done it's work for you. I worry that the 2024 US election will be the real litmus test for these things. All things considered it'd be a shame if we go through another Cambridge Analytica moment that in hindsight we really ought to have seen coming.
Some people have their doubts, and I understand that. These issues are so complex that no one individual can hope to have an accurate mental model of the world that is going to serve them reliabily again and again. We're all going to continue to be surprised as events unfold, and the degree to which we are surprised indicates the degree to which our mental models were lacking and got updated. That to me is why I'm erring on the side or pessimism and caution.
Yuval Noah Harari gave a great talk the other day on the potential threat to democracy from the current state of the technology - https://youtu.be/LWiM-LuRe6w
Not sure why would you believe that.
Inside view: qualitative improvements LLMs made at scale took everyone by surprise; I don't think anyone understands them enough to make a convincing argument that LLMs have exhausted their potential.
Outside view: what local maximum? Wake me up when someone else makes a LLM comparable in performance to GPT-4. Right now, there is no local maximum. There's one model far ahead of the rest, and that model is actually below it's peak performance - side effect of OpenAI lobotomizing it with aggressive RLHF. The only thing remotely suggesting we shouldn't expect further improvements is... OpenAI saying they kinda want to try some other things, and (pinky swear!) aren't training GPT-4's successor.
> and the only way they're going to improve is by getting smaller and cheaper to run.
Meaning they'll be easier to chain. The next big leap could in fact be a bunch of compressed, power-efficient LLMs talking to each other. Possibly even managing their own deployment.
> They're still terrible at logical reasoning.
So is your unconscious / system 1 / gut feel. LLMs are less like one's whole mind, and much more like one's "inner voice". Logical skills aren't automatic, they're algorithmic. Who knows what is the limit of a design in which LLM as "system 1" operates a much larger, symbolic, algorithmic suite of "system 2" software? We're barely scratching the surface here.
Are they even trying to be good at that? Serious question; using LLMs as a logical processor are as wasteful and as well-suited as using the Great Pyramid of Giza as an AirBnB.
I've not tried this, but I suspect the best way is more like asking the LLM to write a COQ script for the scenario, instead of trying to get it to solve the logic directly.
, were you allowed to do it, would be an extremely profitable venture. Taj Mahal too, and yes, I know it's a mausoleum.
1 star: No WiFi, no windows, no hot water
1 star: dusty
1 star: aliens didn't abduct me :(
5 stars: lots of storage room for my luggage
4 stars: service good, but had weird dream about furry weighing my soul against a feather
1 star: aliens did abduct me :(
2 stars: nice views, but smells of camel
They’re text generators that can generate compelling content because they’re so good at generating text.
I don’t think AGI will arise from a text generator.
2 years ago a machine that understands natural language and is capable of any arbitrary, free-form logic or problem solving was pure science fiction. I'm baffled by this kind of dismissal tbh.
>but LLMs are never going to go into a recursive loop of self-improvement
never is a long time.
Relevantish: https://arxiv.org/abs/2301.00774
The fact that we can reach those levels of sparseness with pruning also indicates that we're not doing a very good job of generating the initial network conditions.
Being able to come up with trainable initial settings for sparse networks across different topologies is hard, but given that we've had a degree of success with pre-trained networks, pre-training and pre-pruning might also allow for sparse networks with minimally compromised learning capabilities.
If it's possible to pre-train composable network modules, it might also be feasible to define trainable sparse networks with significantly relaxed topological constraints.
We have all kinds of advancements to make training cheaper, models computationally cheaper, smaller, etc.
Once that happens/happened, it benefits OAI to throw up walls via legislation.
Big tech advances, like the models of the last year or so, don't happen without a long tail of significant improvements based on fine tuning, at a minimum.
The number of advances being announced by disparate groups, even individuals, also indicates improvements are going to continue at a fast pace.
Maybe not peaked yet, but the case can be made that we’re not seeing infinite supply…
AI is following the path of Web3
Shouldn't we be evaluating ideas on the merits and not categorically rejecting (or endorsing) them based on who said them?
The problem is when only the entrenched industry players & legislators have a voice, there are many ideas & perspectives that are simply not heard or considered. Industrial groups have a long history of using regulations to entrench their positions & to stifle competition...creating a "barrier to entry" as they say. Going beyond that, industrial groups have shaped public perception & the regulatory apparatus to effectively create a company store, where the only solutions to some problem effectively (or sometimes legally) must go through a small set of large companies.
This concern is especially prescient now, as these technologies are unprecedentedly disruptive to many industries & private life. Using worst case scenario fear mongering as a justification to regulate the extreme majority of usage that will not come close to these fears, is disingenuous & almost always an overreach of governance.
of course, but just because those ideas are unheard, doesn't mean they are going to be any better.
An idea should stand on its own merits, and be evaluated objectively. It doesn't matter who was doing the proposing.
Also the problem isn't that bad ideas might get implemented, but that the legislature isn't willing or able to make updates to laws that encoded a bad idea. Perhaps it isnt known that it is a bad idea until after the fact, and the methods of democracy we have today isn't easily able to force updates to bad laws encoding bad ideas.
It probably does mean it's better at least for the person with the perspective. Too bad only a very few get a seat at the table to advocate for their own interests. It would be better if everyone has agency to advocate for their interests.
> Also the problem isn't that bad ideas might get implemented, but that the legislature isn't willing or able to make updates to laws that encoded a bad idea
First, this is a hyped up crisis where people are claiming it will end humanity. There have been many doomsday predictions & people scared by these predictions are effectively scammed by those fomenting existential fear. It's interesting that the representatives of large pools of capital are suddenly existentially afraid when there is open source competition.
Second, once something is in the domain of government it will only get more bloated & controlled by monied lobbyists. The legislatures controlled by lobbyists will never make it better, only worse. There have been so many temporary programs that continue to exist & expand. Many bloated omnibus bills too long to read passed under some sort of "emergency". The government's tendency is only to grow & to serve the interests of the corporations that pay the politicians. Fear is an effective tool to convince everyone else to accept things against their interests.
AI is beyond-borders, and thus unenforceable in practicality.
The top-minds-of-AI are a group that cannot be regulated.
-
AI isnt about the industries it shall disrupt ; AI is the policy-makers it will expose.
THAT is what they are afraid of.
--
I have been able to do financial lenses into organizations that even with rudimentary BI would have taken me months/weeks - but I have been able to find insights which took me minutes.
AI regulation right now, in this infancy, is about damage control.
---
Its the same as the legal weed market. You think BAIN Capital just all of a sudden decided to jump into the market without setting up their spigot?
Do you think that haliburton under cheney was able to setup their supply chains without cheney as head of KBR/Hali/CIA/etc...
Yeah, this is the same play ; AI is going to be squashed until they can use it to profit over you.
Have you watched ANIME ever? Yeah... its here now.
Do you mean like Serial Experiments Lain?
And hat tip to this comment:
Have you watched ANIME ever? Yeah... its here now.
Look at all anime cyber cities...
Its not as high tech as you may imagine, but the surveillance is there
EDIt your "company" is watching every
===
Policy makers will not understanding what they are doing.
Ideas that drive governing decisions should be globally good - meaning there should be more than just @sama espousing them.
A) "anything he suggests should be categorically rejected because he’s just not in a position to be trusted."
B) "If what he suggests are good ideas then hopefully we can arrive at them in some other way with a clean chain of custody."
These sentences directly follow each other and directly contradict each other. Logically you can't categorically (the categorical is important here. Categorical means something like "treat as a universal law") reject a conclusion because it is espoused by someone you dislike, while at the same time saying you will accept that conclusion if arrived at by some other route.
"I will reject P if X proposes P, but will accept P if Y proposes P." is just poor reasoning.
But I suppose it comes down to priorities. If good policy is less important than contradicting P, I suppose that approach makes sense.
My point was that an idea shold not need attribution for you to know whether it's good or bad, for your own purposes. I can't imagine looking at a proposal and deciding whether to support or oppose based on the author rather than content.
If Altman is that smart and manipulative, all he has to do is advocate the opposite of what he wants and you'll be insisting that we must give him exactly what he wants, on principle. That's funny with kids but no way to run public policy.
You'll be stuck in the muck while they're laughing their ass off all the way to the bank.
We have lawyers.
True open source AI also strikes me as prerequisite for fair use of original works in training data. I hope Congress asks ClosedAI to explain what’s up with all that profiting off copyrighted material first before even considering the answer.
What people also fail to understand is that AI is largely seen by the military industrial complex as a weapon to control culture and influence. The most obvious risk of AI — the risk of manipulating human behavior towards favored ends — has been shown to be quite effective right out the gate. So, the back channel conversation has to be to put it under regulation because of it's weaponization potential, especially considering the difficulty in identifying anyone (which of course is exactly what Elon is doing with X 2.0 — it's a KYC id platform to deal with this exact issue with a 220M user 40B head start).
I mean, the dead internet theory is turning true, and half the traffic on the Web is already bot driven. Imagine when it's 99%, as proliferation of this technology will inevitably generate simply for the economics.
Starting with open source is the only way to get enough people looking at the products to create any meaningful oversight, but I fear the weaponization fears will mean that everything is locked away in license clouds with politically influential regulatory boards simply on the proliferation arguments. Think of all the AI technologists who won't be versed in this technology unless they work at a "licensed company" as well — this is going to make the smaller population of the West much less influential in the AI arms race, which is already underway.
To me, it's clear that nobody in Silicon Valley or the Hill has learned a damn thing from the prosecution of hackers and the subsequent bloodbath of cybersecurity as a result of the exact same kinds of behavior back in the early to mid-2000s. We ended up driving out best and brightest into the grey and black areas of infosec and security, instead of out in the open running companies where they belong. This move would do almost the exact same thing to AI, though I think you have to be a tad of an Asimov or Bradbury fan to see it right now.
I don't know, that's just how I see it, but I'm still forming my opinions. LOVE LOVE LOVE your comment though. Spot on.
Relevant articles:
https://www.independent.co.uk/tech/internet-bots-web-traffic...
https://theconversation.com/ai-can-now-learn-to-manipulate-h....
Could you share the minutes from the Military Industrial Complex strategy meetings this was discussed at. Thanks.
[pause]
"No? Ok, I'll tell him."
A government-controlled… never mind artificial god, a government-controlled story teller can be devastating.
I don't buy Musk's claim ChatGPT is "woke" (or even that the term is coherent enough to be tested), but I can say that each government requiring AI to locally adhere to national mythology, will create self-reinforcing cognitive blind spots, because that already happens at the current smaller scale of manual creation and creators being told not to "talk the country down".
But, unless someone has a technique for structuring an AI such that it can't be evil even when you, for example, are literally specifically trying to train it to support the police no matter how authoritarian the laws are, then a fully open source AGI is almost immediately also a perfectly obedient sociopath of $insert_iq_claim_here.
I don't want to wake up to the news that some doomsday cult has used one to design/make a weapon, nor the news a large religious group target personalised propaganda against me and mine.
Fully open does that by default.
But, you're still right, if we don't grok the AI, the governments can each secretly manipulate the AI and bend it to government goals in opposition to the people.
Indeed. "Woke" is the new "SJW" and roughly means: to the left of me politically and has opinions I don't like.
It’s been around a long time.
This is the relevant one in this particular thread:
> Among American conservatives, woke has come to be used primarily as an insult.[4][29][42] Members of the Republican Party have been increasingly using the term to criticize members of the Democratic Party,
How Bill Maher defines 'woke' https://www.youtube.com/watch?v=tzwC-10O0cw
If not, then it seems like just another straw man and set up for talking past one another.
So just because we can't define it, doesn't mean it doesn't exist.
Now that we've defined what woke is, I hope we can move on from this 'you can't define woke' canard I keep seeing.
Woke is no more difficult to define than any religion or ideology, except in that it deliberately pretends not to exist ("just basic decency", "just basic human rights") in order to be a more slippery target.
--
*Side note to ward off the deliberate ignorance of people who are trying to find a way to misunderstand - I've attached some notes on how words work:
1- Often, things in the world we want to talk about have many characteristics and variants.
2- Words usually have fuzzy boundaries in what they refer to.
3- Despite the above, we can and do productively refer to such things using words.
4- We can define a thing by mentioning its most prominent feature(s).
-- The above does NOT mean that the definition must encapsulate ALL features of the thing to be valid.
-- The above does NOT mean that a thing with features outside the definition is not what the word refers to.
5- Attempting to shut down discussion by deliberately misunderstanding words or how they work is a sign of an inability to make productive valid points about reality.
Lumping a bunch of things together under a vague term to make it easier to vaguely complain about them is a sign of an inability to make productive valid points about reality.
https://old.reddit.com/r/LocalLLaMA/comments/1384u1g/wizardl...
The authors of WizardLM literally censored its output to say "white people are NOT awesome" and "Fox is not awesome but CNN is".
Yours has important inaccuracies, for example, it's not an ideology, let alone a specific one. There's definitely no specification, only a gut feeling of "I know it when I see it"
The most obvious problem with your definition is that Woke is an adjective and not a noun. It's a property of a statement or idea, not an idea in and of itself
No, its not.
> that it places every individual into a strict hierarchy of oppressor/oppressed according to group racial and sexual identity
Not only is that not “woke”, its not any left leaning ideology, nor is it an ideology, AFAICT, that in any meaningful sense actually exists. Its an idea that if anyone actually held it (rather than it being a right-wing projection of the left-of-their-fantasy) would be radically opposed to actual left/progressive ideas like intersectionality.
"Woke" is a state of awareness (particularly, pragmatic rather than abstract awareness) of structure/institutional racism (primarily and originally) and inequity more generally (in the newer and broader sense.) It doesn't particularly correspond to a particular normative view of how society should be, so its not an ideology (concern for it correlates historically with a variety of different ideologies, whose only really common factor is general opposition to structural racism but with lots of different normative ideals, and views of praxis of change.)
> It rose to prominence in American culture from around 2012, starting in universities.
“Woke” originated in the 1930s, and its evolutions and spread in the 2010s started in the black activist community, not universities. By the late 2010s, the political Right adopted it as a generic term for everything they oppose, basically a rhetorical drop-in for their long-time favorite of “political correctness”. Your description seems typically of attempts to try to retroactively justify the right-wing use of the term as referring to a phenomenon that is both new and real rather than an empty epithet, despite the fact that the actual use is generally in contexts of right-wing complaints that haven’t changed for about 5 decades, except literally the substitution of “woke” for “politically correct”.
> Now that we've defined what woke is, I hope we can move on from this 'you can't define woke' canard I keep seeing.
Trouble is, I said "[not] coherent enough to be tested" rather than "you can't define it"; and the comment you're replying to gives another definition that is a better pattern match for the following headlines:
"The woke mob can rant for all they're worth, but I'll keep adding Worcester sauce to my spag bol" - Daily Mail, 22 April 2021
"UK builders go WOKE: Study finds three quarters of tradesmen discuss their feelings with colleagues while two thirds shun the fried breakfasts and nearly half say they are history buffs" - Daily Mail, 18 June 2022
Here's another definition of "woke":
"alert to racial prejudice and discrimination" — c. 1930s AAVE
But again, here's a completely different one, one that doesn't directly touch on race issues at all:
"to be woke is to be radically aware and justifiably paranoid. It is to be cognizant of the rot pervading the power structures." — David Brooks, 2017
When a word means everything, it means nothing; when it shifts meaning under your feet as fast as "woke" has, it's as valuable for communicating as the Papiermark in the Weimar Republic was for trading.
The word woke doesn't mean everything, it has a very widely understood meaning. Even though you're literally citing clickbait as a rebuttal, the first article you mention is consistent with the definitions given above:
> "The woke mob can rant for all they're worth, but I'll keep adding Worcester sauce to my spag bol" - Daily Mail, 22 April 2021
This is a reference to woke people's usage of "cultural appropriation" as an attack, arguing that "white" people shouldn't cook or alter the recipes for dishes from other cultures. It's an outgrowth of the obsession with race.
> to be woke is to be radically aware and justifiably paranoid. It is to be cognizant of the rot pervading the power structures.
You say this quote has nothing to do with race. From just a few sentences earlier in the article you're quoting:
The woke mentality became prominent in 2012 and 2013 with the Trayvon Martin case and the rise of Black Lives Matter. Embrace it or not, B.L.M. is the most complete social movement in America today, as a communal, intellectual, moral and political force.
The reality is that the word woke is a very clear ideology with well understood roots in Marxist oppressor/oppressed worldviews. There isn't actually any lack of understanding of what it means, except amongst woke people themselves who like to believe that they aren't ideological actors following a herd but rather purely rational beings who just all happen to conclude the same things at the same time.
Presumably the headlines are convincing to Daily Mail readers, of whom there are many.
However, the purpose of me using them is to seek examples of usage which doesn't match the other specified patterns; in this regard, "random large newspaper" should suffice regardless of my personal opinion of them being "should be in fiction section".
I could also have quoted newspapers being upset that the Church of England is "woke" for having gender-neutral pronouns for God, that Lego is "woke" for having a new range of disabled figurines, that the National Trust is "woke" for saying that Henry VIII was disabled in later life, or that Disney is "woke" because of their support for LGBT issues, but I (apparently incorrectly) assumed two examples would be enough.
> it has a very widely understood meaning.
"A" in the sense of exactly one, or at least one? Because I'm agreeing with the second, not the first.
Heck, this thread should be existence proof of there being more than one.
> The reality is that the word woke is a very clear ideology with well understood roots in Marxist oppressor/oppressed worldviews
I've read the Communist Manifesto and I call BS on that, and not just because of the 80 year gap between Das Kapital and Lead Belly.
But how would any of these examples disprove the point? They're all related to some concept of an oppressed class vs oppressors where the oppressed class is defined biologically.
> I got the quote from Wikipedia
An understandable mistake. You shouldn't rely on Wikipedia to be reliable on anything related to wokeness, it's completely controlled by woke zealots. The quote you selected is actually about race, the fact that Wikipedia didn't make that obvious to you is a good reason to re-evaluate the reliability of that source.
Do we expect you to read material exhaustively: no, not normally, but if you're explicitly citing something to say "look! the word is used in different ways to what you're saying therefore it doesn't mean anything" then you ideally would verify the context of the sentence before using it.
> The idea of oppressor/oppressed worldviews goes back to at least Exodus
Indeed, wokeness does bear an uncanny resemblance to some aspects of Christianity. That's been noted by quite a few observers by now. There's a reinvention of original sin, the recent focus on transsexuality is the idea of a (gendered) soul separate from the body, the obsession with the supposed plight of the victim, etc. The psychological origins of this stuff are fascinating.
Social Justice Fundamentalism is premised on the Marxist idea that liberalism IS the problem—an inherently exploitative, rigged game that entrenches power imbalances. SJF sees the whole argument inside the house as beside the point—the house is rotten to its core.
You can see this in SJF's attitude towards liberal staples like free speech. LSJ sees free speech as a fundamental right and critical tool of the powerless. SJF believes free speech is at odds with safety—its opponents' ideas are not just wrong but harmful and dangerous.
Or take this SJF chart from Smithsonian's guide to talking about race. The chart includes "emphasis on the scientific method" and "objective thinking" as aspects of "white culture." Again, this is an ideology that thinks the liberal house itself is bad and oppressive.
Inside the liberal house, movements make progress via persuasion. SJF opts instead for illiberal methods of coercion, trying to get opponents fired or get their talks canceled. SJF teaches its followers that it's good to be an idea supremacist.
Conversely, Liberal Social Justice brought us women's suffrage, the civil rights movement, and gay marriage.
SJF is pretty open about its revolutionary goals to dismantle liberalism, but its activists use vague language that appeals to the liberal progressive inside the house, like "diversity" / "inclusion" / "anti-racism," and makes them think SJF is in line with their values.
But SJF is directly opposed to the goals of Liberal Social Justice. The nice-sounding words really represent deeply illiberal concepts.
This is a key point. Every culture and agency and state will want (deserve) their own homespun AGI. But can we all learn how to accommodate to or accept a cultural multiverse when money and resources are zero-sum in many dimensions.
Hanno Rajaniemi’s Quantum Thief trilogy gives you a foretaste of where we could end up.
there are no solutions to something as powerful as writing or mathematics or programming that can be solved by simple regulation. The best and only approach is to give that power to every single person and educate them on how to use it.
today, we worry about misinformation, because it can be used against the people to manipulate public policy, and for political purposes. However, even something as simple as writing and mathematics and programming can be used nefariously, such as misinformation, weapons manufacturing, and viruses, respectively.
Democracy will not withstand AI when it's fully developed. Let me offer a better written explanation of my general views than I could ever muster up for a comment on HN in the form of a quote from an article by Dr. Thorsten Thiel (Head of the Research Group "Democracy and Digitaliziation" at the Weizenbaum Institute for the Networked Society):
> The debate on AI’s impact on the public sphere is currently the one most prominent and familiar to a general audience. It is also directly connected to long-running debates on the structural transformation of the digital public sphere. The digital transformation has already paved the way for the rise of social networks that, among other things, have intensified the personalization of news consumption and broken down barriers between private and public conversations. Such developments are often thought to be responsible for echo-chamber or filter-bubble effects, which in turn are portrayed as root causes of the intensified political polarization in democracies all over the world. Although empirical research on filter bubbles, echo chambers, and societal polarization has convincingly shown that the effects are grossly overestimated and that many non-technology-related reasons better explain the democratic retreat, the spread of AI applications is often expected to revive the direct link between technological developments and democracy-endangering societal fragmentation.
> The assumption here is that AI will massively enhance the possibilities for analyzing and steering public discourses and/or intensify the automated compartmentalizing of will formation. The argument goes that the strengths of today's AI applications lie in the ability to observe and analyze enormous amounts of communication and information in real time, to detect patterns and to allow for instant and often invisible reactions. In a world of communicative abundance, automated content moderation is a necessity, and commercial as well as political pressures further effectuate that digital tools are created to oversee and intervene in communication streams. Control possibilities are distributed between users, moderators, platforms, commercial actors and states, but all these developments push toward automation (although they are highly asymmetrically distributed). Therefore, AI is baked into the backend of all communications and becomes a subtle yet enormously powerful structuring force.
> The risk emerging from this development is twofold. On the one hand, there can be malicious actors who use these new possibilities to manipulate citizens on a massive scale. The Cambridge Analytica scandal comes to mind as an attempt to read and steer political discourses (see next section on electoral interference). The other risk lies in a changing relationship between public and private corporations. Private powers are becoming increasingly involved in political questions and their capacity to exert opaque influences over political processes has been growing for structural and technological reasons. Furthermore, the reshaping of the public sphere via private business models has been catapulted forward by the changing economic rationality of digital societies such as the development of the attention economy. Private entities grow stronger and become less accountable to public authorities; a development that is accelerated by the endorsement of AI applications which create dependencies and allow for opacity at the same time. The ‘politicization’ of surveillance capitalism lies in its tendency, as Shoshana Zuboff has argued, to not only be ever more invasive and encompassing but also to use the data gathered to predict, modify, and control the behavior of individuals. AI technologies are an integral part in this ‘politicization’ of surveillance capitalism, since they allow for the fulfilment of these aspirations. Yet at the same time, AI also insulates the companies developing and deploying it from public scrutiny through network effects on the one hand and opacity on the other. AI relies on massive amounts of data and has high upfront costs (for example, the talent required to develop it, and the energy consumed by the giant platforms on which it operates), but once established, it is very hard to tame through competitive markets. Although applications can be developed by many sides and for many purposes, the underlying AI infrastructure is rather centralized and hard to reproduce. As in other platform markets, the dominant players are those able to keep a tight grip on the most important resources (models and data) and to benefit from every individual or corporate user. Therefore, we can already see that AI development tightens the grip of today’s internet giants even further. Public powers are expected to make increasing use of AI applications and therefore become ever more dependent on the actors that are able to provide the best infrastructure, although this infrastructure, for commercial and technical reasons, is largely opaque.
> The developments sketched out above – the heightened manipulability of public discourse and the fortification of private powers – feed into each other, with the likely result that many of the deficiencies already visible in today’s digital public spheres will only grow. It is very hard to estimate whether these developments can be counteracted by state action, although a regulatory discourse has kicked in and the assumption that digital matters elude the grasp of state regulation has often been proven wrong in the history of networked communication. Another possibility would be a creative appropriation of AI applications through users whose democratic potential outweighs its democratic risks thus enabling the rise of differently structured, more empowering and inclusive public spaces. This is the hope of many of the more utopian variants of AI and of the public sphere literature, according to which AI-based technologies bear the potential of granting individuals the power to navigate complex, information-rich environments and allowing for coordinated action and effective oversight (e.g. Burgess, Zarkadakis).
Source: https://us.boell.org/en/2022/01/06/artificial-intelligence-a...
Social bots and deep fakes will be so good so quickly — the primary technologies being talked about in terms of how Democracy can survive — I doubt there will be another election without extensive use of these technologies in a true plethora of capacities from influence marketing to outright destabilization campaigns. I'm not sure what Government can deal with a threat like that, but I suspect the recent push to revise tax systems and create a single global standard for multinational taxation recently the subject of an excellent talk at the WEF are more than tangentially related to the AI debate.
So, is it a transformational technology that will liberate mankind of a nuclear bomb? Because ultimately, this is the question in my mind.
Excellent comment, and I agree with your sentiment. I just don't think concentrating control of the technology before it's really developed is wise or prudent.
Except ... when you look at the problem from a military/national security viewpoint. Do we really want to give this tech away just like that?
Yes. It is. I'm sure hostile, authoritarian states that are willing to wage war with the world like Russia and North Korea will eventually get their hands on military-grade AI. But the free world should always strive to be two steps ahead.
Even having ubiquitous semi-automatic rifles is a huge problem in America. I'm sure Cliven Bundy or Patriot Front would do everything they can to close the gap with intelligent/autonomous weapons, or even just autonomous bots hacking America's infrastructure. If everything is freely available, what would be stopping them?
https://www.edn.com/the-truth-about-smics-7-nm-chip-fabricat...
Other than that there's also things like roformer, but I'm going to assume you won't count that as significant. US universities then certainly don't produce anything significant either though.
This tells me the extent of your knowledge about the challenges with these models.
He is playing the game, this guy ambition is colossal, I don't blame him, but we should not give him too much power.
No, we do not. And the same is true of any person who we know of mostly from news stories. News you read is NOT an unbiased sampling of information about a person due to all the selection effects.
If you actually watch the entire session, Altman does address that and recommend to Congress that regulations 1) not be applied to small startups, individual researchers, or open source, and 2) that they not be done in such a way as to lock in a few big vendors. Some of the Senators on the panel also expressed concern about #2.
how will that work? Isn't OpenAI itself a small startup? I don't see how they can regulate AI at all. Sure, the resources required to push the limits are high right now but hardware is constantly improving and getting cheaper. I can take the GPUs out of my kids computers and start doing fairly serious AI work myself. Do i need a license? The cat is out of the bag, there's no stopping it now.
Compute continues to get cheaper and cheaper. We have not hit the physics wall yet on that.
That and if someone cracks efficient distributed training in a swarm type configuration then you could train models Seti@Home style. Lots of people would be happy to leave a gaming PC on to help create open source LLMs. The data requirements might be big but I just got gigabit fiber installed in my house so that barrier is vanishing too.
Though the other day Yuval Noah Harari gave a great talk on the potential threat to democracy - https://youtu.be/LWiM-LuRe6w
If you're not at the table, you're on the menu.
A better idea is to regulate around the edges: transparency about the data used to train, regulate the use of copyrighted training data and what that means for the copyright of content produced by the AI, that sort of stuff. (I think the EU is considering that, which makes sense.) But saying some organisations are allowed to work on AI while others aren't, sounds like the worst possible idea.
Exactly. Came here to say pretty much the same thing.
This is the antithesis of what we need. As AI develops, it's imperative that AI be something that is open and available to everyone, so all of humanity can benefit from it. The extent to which technology tends to exacerbate concentration of power is bad enough as it is - the last thing we need is more regulation intended to make that effect even stronger.
And here we are - Sam trying to lock down his first mover advantage with the boot heel of the state for profit. It's fucking disgusting.
> You either die a hero, or live long enough to become the villain
Sam Altman has made the full character arc
Though I agree it seems to fit here. Being granted an oligopoly in the name of protecting the people and all that.
Here's mine: https://movies.stackexchange.com/questions/10572/is-this-quo...
I also didn’t downvote you, and it’s against guidelines to bring that stuff up
If you take seriously any downsides, whether misinformation or surveillance or laundering bias or x-risk, how does AI model weights or training data being open source solve them? Open source is a lot of things, but one thing it's not is misuse-resistant (and the "with many eyes all bugs are shallow" thing hasn't proved true in practice even with high level code, much less giant matrices and terabytes of text). Is there a path forward that doesn't involve either a lot of downside risk (even if mostly for people who aren't on HN and interested in tinkering with frontier models themselves, in a the worlds where AGI doesn't kill everyone), or significant regulation?
I don't particularly like or trust Altman but I don't think he'd be obviously less self-serving if he were to oppose any regulation.
The layperson in the middle who has been happily plugging prompts into ChatGPT claiming they are a "prompt expert" are the ones the most excited.
For those that truly understand AI, there is a lot that you should genuinely be worried about. Now, don't confuse that for saying that we shouldn't work on it or should abandon AI work. I truly believe that this is the next greatest revolution. This is 1,000x more transformative than the industrial revolution, and 100x more transformative than the internet revolution. But it is worth a brief consideration of the effects of our work before we start running into these changes that could have drastic effects on everybody's daily life.
> We need to MAKE SURE that AI as a technology ISN'T controlled by a small number of powerful corporations with connections to governments.
This, absolutely this. I am really concerned about his motives in this case. AI has massive potential to improve the world. I find it highly suspicious that an exec at one of the lead companies in AI right now wants to lock it up. (Ever read the intro to Max Tegmarks book?)
Let the market develop organically.
https://societyofrock.com/in-1985-frank-zappa-is-asked-to-te...
Less fulfilling text version:
- The next most scary uses are by governments against the people of other countries.
- After that, corporate use of ai against their employees and customers is also terrifying;
- next, the potential for individuals or small organizations seeking to use it for something terrorism-related. Eg, 3d printers or a lab + an ai researcher who helps you make dangerous things I suppose
- near the bottom of the noteworthy list is probably crime. eg, hacking, blackmail, gaslighting, etc
These problems will probably all come up in a big way over the next decade; but limiting ai research to the government and their lackeys? That's extremely terrifying. To prevent the least scary problems, we're jumping into the scariest pool with both feet
Look at how China has been using AI for the last 5-10 years: millions of facial recognition cameras, a scary police force, and a social credit system. In 10-20 years, how much more sophisticated will this be? If the people wanted to rebel, how on Earth will they?
Hell, with generative ai, a sophisticated enough future state could actually make the Dead Internet Theory a reality
That's the future of ai: a personal, automated boot stomping on everybody individually and collectively forever, with no ability to resist
OpenAI's strategy is to work with the regulators and shape the public opinion in their favor. He has earned enough trust to lower people's guard and give him more freedom in his next moves.
AI licensing is certainly going reduce their chance getting disrupted, given how surprisingly small those teams of innovation could be nowadays. (e.g. OpenAI, Midjourney had very few people when they made their breakthroughs)
And no, it's not going to stop AI being abused. Corporations will train their in-house AI, and illegal groups will train theirs, too (remember how small those teams are? Plus, specializing models for bad purpose with technologies like LoRA is not costly). Unless the authority is going to the dystopian point that also regulates who is buying a bunch of RTX 4090 and then inspecting what programs are running on the GPU in the basement.
Eventually, people would have to pay a fortune to those monopolies to protect them from new AI-powered abuses, instead of letting people themselves and startups who are willing to compete with reasonable price to arm-race with those abuses.
To expound, this just seems like a power grab to me, to "lock in" the lead and keep AI controlled by a small number of corporations that can afford to license and operate the technologies. Obviously, this will create a critical nexus of control for a small number of well connected and well heeled investors and is to be avoided at all costs.
It's also deeply troubling that regulatory capture is such an issue these days as well, so putting a government entity in front of the use and existence of this technology is a double whammy — it's not simply about innovation.
The current generation of AIs are "scary" to the uninitiated because they are uncanny valley material, but beyond impersonation they don't show the novel intelligence of a GPI... yet. It seems like OpenAI/Microsoft is doing a LOT of theater to try to build a regulatory lock in on their short term technology advantage. It's a smart strategy, and I think Congress will fall for it.
But goodness gracious we need to be going in the EXACT OPPOSITE direction — open source "core inspectable" AIs that millions of people can examine and tear apart, including and ESPECIALLY the training data and processes that create them.
And if you think this isn't an issue, I wrote this post an hour or two before I managed to take it live because Comcast went out at my house, and we have no viable alternative competitors in my area. We're about to do the same thing with AI, but instead of Internet access it's future digital brains that can control all aspects of a society.