When you agreed to the terms of the various sites you use every day, you allowed the companies access to your output which they will use to train their models. It’s too late.
You can show people who don't understand it how it just produces statistical word salad, and it is not an "intelligence" by any reasonable definition. An easy way to do this is to ask to to do a simple math problem and it will give the wrong answer. Another easy way is to ask it something where a SME would easily see the errors. For example if you have a friend who is a doctor give it some medical questions and they'll probably see lots of errors. Same for a lawyer, or a farmer, or whatever. Essentially do whatever you can to counter OpenAI's lies and hype.
Familiarize your audience with the story "Liar!" from Asimov's I, Robot anthology, and then show them how GPT-3 does the same thing as the robot from the story: guess what its user wants to hear and tell them that (albeit with statistical modelling instead of mind reading).
If new research comes out that most human cognition tasks (read middle class jobs) requires only n joint distributions to model and once we scaled up to n, are you still going to insist on hiding your head in the sand? Shaking your fists at the word salads won't bring those jobs back.
Taking Gebru's word seriously here is reason to discount the entire argument.
Gebru is a part of a clique that is devoted to an entirely different set of ideas around AI Safety.
The two basic movements here are:
1. Be very scared of AI development, then do capabilities research but feel bad about it.
2. Demand AI research comes with locks in them that make sure the AI can't write a tweet that would make a San Franciscan Activist uncomfortable. Achieve very little in terms of manipulating AI, but get lots of book deals.
Neither of these groups should really be trusted for hard data about the other, and more importantly - missing that piece indicates that the author is not even attempting to convince the unconvinced.
Exactly, I've read and listened to a lot of people being overly dismissive of chat gpt. And I've also used it. It's obviously not perfect but it clearly does lots of cool and useful things that go a bit beyond just echoing what wikipedia and stackoverflow say.
A pattern that I see a lot with the debate around this topic is people leaning a lot on their reputations rather than hard arguments. I was listening to a few podcasts on this topic and they all featured a dr of this or that with a lot of books pontificating about their version of the truth and how others were clearly wrong. A lot of these 'experts' are parroting each other as well (chat gpt would do a great job of making their arguments for them). You get a lot of "he said that she said" type argumentation and name dropping. Not going to name names here but I just don't find this line of argumentation very convincing or productive. There's a lot of extrapolation and interpreting that is done by these people that is based on a basis that just isn't that great.
ChatGpt obviously happened and that seems to excite, worry, or upset lots of people. However, that cat is out of the bag now. OpenAI's results are being replicated by countless third parties. No amount of legislation, self censuring, pleading, magical thinking, etc. is going to change that. So, I don't buy into the whole Luddite attitude. I just don't think it is productive to think like that or insist that others think like that. The only debate worth having is how to improve the technology further and how to deal with the inevitable abuse that will happen.
Also, this is obviously not an AGI yet. Whatever that is. And obviously some people are trying to inch closer to that goal. And there's a lot of debate about what is the right approach for getting there with many competing points of view. That's a much more interesting debate. Interesting time to be alive. So much progress in so little time. And I'm not even 50 yet. The debate that this cannot be allowed to happen or is impossible is a lot less interesting to me.
My favorite part of OpenAI chat is that there are no videos, no popups asking me to sign up for a newsletter, no warnings to enable cookies, no hiding content behind interstitial ads on scroll.
I will gladly pay good money for a web experience and API that just answers my questions.
I don’t know if you were around then but early google results were also without ads (or at least unassuming ones), this is just the way early tech adoption works.
Additionally: openAI actually wants prompts from humans; these tools are essentially data gathering part of their ongoing research.
A documented, opiniated, non-hype biased post about AIs. I've lived the 80's AI hype wave and the fall that followed. AFAICT we're on the same train here.
"Any sufficiently advanced technology is indistinguishable from magic" (Arthur C Clarke) and we're waiting for the masses to recognize what magic sparkles are : dust (ok, weighted barycenters of a huge database)
Side note : I often wonder how *GPT can deal with self-reference, like "What would you think of you if you were me ?" or "tell me something you were not programmed to think is plausible"
> I've lived the 80's AI hype wave and the fall that followed. AFAICT we're on the same train here.
Really? In the 80s they were trying to teach software to recognize handwritten numbers. Today, corporations are tightly controlling public access to large language models so they aren't used to manipulate elections and unhinge society.
That doesn't quite seem comparable to me. If anything, models like ChatGPT are underhyped because tech journalists mostly focus on stupid mistakes those models make in some situations.
Excellent point. I think it’s critical to recognize that as AI tech is getting more advanced, we need to improve the mechanisms that enforce transparency and trustability. It starts with training data. Just because something is technically reachable online, doesn’t make it OK for it to be used for AI training. A person or organization should be able to opt into having their data used, of course, but it’s not a “use first, ask to remove later”.
Then who is watching the watchers? Models being used to generate content or make decisions that have a real-world impact on actual humans need to be open to audit and validation that they don’t repeat the bias and social limitations of the people who create those models. This is not a new concept (think algorithms for insurance coverage calculations etc).
So yes, a toy chatbot to play around with - sure. A fancy AI model, sucking personal data without consent and being deployed all around us by closed organizations with dubious funding - big no. Is AI ready to offer scientific or academic value - no, and even if it was a yes, it will need to stand to human scrutiny for a long time before being acceptable “as is”. Kids in schools need to be thought how to recognise and work with AI generated content. Just like dealing with fake news, it requires critical thinking and the ability to analyse information.
My concern is who is watching the ethicists who are making decisions and regulations "for our own good". All the discussion about AI that i can read about is clearly a power play about who will get to set the rules (and rent-seek) first
Folks who specialize in issues arising from algorithmic bias, ethics, and techniques for gathering data and scraping, data lineage etc. are uniquely qualified to support the ethical perspective.
This is not competition. Politicians/corporates/religious leaders/influencers and such want it to be a competition, so they can reinforce their base of power and influence. That's why we must develop tools to guard against such influences.
This seems naive and inviting exactly the kind of bias we need to avoid. For example, you say “democratic” which means different things to different groups of people.
Transparency and trustability are precisely the tools that once installed can allow the public to peer into the workings of an AI system, it's models and training data. To achieve that, we need the cooperation of a wide variety of specialists in domains like ethics, computer science, law etc. Also, the right choices must be made while creating that AI system (e.g. in regard to data collection, diversity of the sample data etc).
> Transparency and trustability are precisely the tools
I don't get the point of transparency if "we the people" dont have a say on the politics of AI that affects our life. AI is too important to be left to a cabal of "specialists"
> AI is too important to be left to a cabal of "specialists"
You do realize you're arguing against yourself, right? "Cabal of specialists" could also apply to the people who work on AI. For profit. Information asymmetry is well known to be the main cause of market failure. How is that cabal alone better than that plus people with other specialties such as ethics, plus everyone able to watch them all and decide for themselves whether that aligns with their own (or society's) interests?
I'm not sure who is more naive: The people who believe that this technology is the solution to every problem, or the people who believe that it is somehow possible to put this genie back into the box.
AI will change everything. Not in the distant future when it becomes sentient, but in the next 5-10 years.
I'm not sure the paper's authors are very diverse as they claim, they seem cut from the same cloth to me.
ChatGPT may spit out word salad, and be trained on inappropriate input, but there is no doubting the success of some of these models, eg at playing the game of Go, predicting protein folding, etc, etc.
There's obviously something still missing, but the hype is warranted IMHO, these model represent a major advance compared to even just a few years ago.
ChatGPT is like a super charged rubber duck. It often doesn’t give me the whole answer or the right answer. But it unblocks me and gives me ideas to keep moving.
Ah yes, academics, the most unbiased angels of us all…
There are good arguments in this article, buried in a mire of self righteous indignation.
Better education on viewing any output by a model with more critical eyes - which really is today a basic requirement for parsing any online information - is a fine bill to die on.
That then goes in hand with viewing with suspicious eyes the motivations of companies that are commercialising this technology, which is nowhere near a new sentiment.
However better recognition needs to be given to what obviously fuels the hype being railed against: the utility of this tool, even if it fails to work adequately most of the time. YMMV depending on your personal experience on the queried topics.
The article does indeed sound like it’s just preaching to the choir.
40 comments
[ 3.2 ms ] story [ 151 ms ] threadI don't think you realize how bad it is out there. gestures.
Could you be more specific, to make discussion possible?
Gebru is a part of a clique that is devoted to an entirely different set of ideas around AI Safety.
The two basic movements here are: 1. Be very scared of AI development, then do capabilities research but feel bad about it. 2. Demand AI research comes with locks in them that make sure the AI can't write a tweet that would make a San Franciscan Activist uncomfortable. Achieve very little in terms of manipulating AI, but get lots of book deals.
Neither of these groups should really be trusted for hard data about the other, and more importantly - missing that piece indicates that the author is not even attempting to convince the unconvinced.
A pattern that I see a lot with the debate around this topic is people leaning a lot on their reputations rather than hard arguments. I was listening to a few podcasts on this topic and they all featured a dr of this or that with a lot of books pontificating about their version of the truth and how others were clearly wrong. A lot of these 'experts' are parroting each other as well (chat gpt would do a great job of making their arguments for them). You get a lot of "he said that she said" type argumentation and name dropping. Not going to name names here but I just don't find this line of argumentation very convincing or productive. There's a lot of extrapolation and interpreting that is done by these people that is based on a basis that just isn't that great.
ChatGpt obviously happened and that seems to excite, worry, or upset lots of people. However, that cat is out of the bag now. OpenAI's results are being replicated by countless third parties. No amount of legislation, self censuring, pleading, magical thinking, etc. is going to change that. So, I don't buy into the whole Luddite attitude. I just don't think it is productive to think like that or insist that others think like that. The only debate worth having is how to improve the technology further and how to deal with the inevitable abuse that will happen.
Also, this is obviously not an AGI yet. Whatever that is. And obviously some people are trying to inch closer to that goal. And there's a lot of debate about what is the right approach for getting there with many competing points of view. That's a much more interesting debate. Interesting time to be alive. So much progress in so little time. And I'm not even 50 yet. The debate that this cannot be allowed to happen or is impossible is a lot less interesting to me.
The hype is, in large part, real to me. GPTs, but also Whisper or Stable Diffusion/DALLE solve real problems, way better than previous tools could.
I will gladly pay good money for a web experience and API that just answers my questions.
Additionally: openAI actually wants prompts from humans; these tools are essentially data gathering part of their ongoing research.
A documented, opiniated, non-hype biased post about AIs. I've lived the 80's AI hype wave and the fall that followed. AFAICT we're on the same train here.
"Any sufficiently advanced technology is indistinguishable from magic" (Arthur C Clarke) and we're waiting for the masses to recognize what magic sparkles are : dust (ok, weighted barycenters of a huge database)
Side note : I often wonder how *GPT can deal with self-reference, like "What would you think of you if you were me ?" or "tell me something you were not programmed to think is plausible"
Really? In the 80s they were trying to teach software to recognize handwritten numbers. Today, corporations are tightly controlling public access to large language models so they aren't used to manipulate elections and unhinge society.
That doesn't quite seem comparable to me. If anything, models like ChatGPT are underhyped because tech journalists mostly focus on stupid mistakes those models make in some situations.
Then who is watching the watchers? Models being used to generate content or make decisions that have a real-world impact on actual humans need to be open to audit and validation that they don’t repeat the bias and social limitations of the people who create those models. This is not a new concept (think algorithms for insurance coverage calculations etc).
So yes, a toy chatbot to play around with - sure. A fancy AI model, sucking personal data without consent and being deployed all around us by closed organizations with dubious funding - big no. Is AI ready to offer scientific or academic value - no, and even if it was a yes, it will need to stand to human scrutiny for a long time before being acceptable “as is”. Kids in schools need to be thought how to recognise and work with AI generated content. Just like dealing with fake news, it requires critical thinking and the ability to analyse information.
My concern is who is watching the ethicists who are making decisions and regulations "for our own good". All the discussion about AI that i can read about is clearly a power play about who will get to set the rules (and rent-seek) first
This is not competition. Politicians/corporates/religious leaders/influencers and such want it to be a competition, so they can reinforce their base of power and influence. That's why we must develop tools to guard against such influences.
Transparency and trustability are precisely the tools that once installed can allow the public to peer into the workings of an AI system, it's models and training data. To achieve that, we need the cooperation of a wide variety of specialists in domains like ethics, computer science, law etc. Also, the right choices must be made while creating that AI system (e.g. in regard to data collection, diversity of the sample data etc).
I don't get the point of transparency if "we the people" dont have a say on the politics of AI that affects our life. AI is too important to be left to a cabal of "specialists"
You do realize you're arguing against yourself, right? "Cabal of specialists" could also apply to the people who work on AI. For profit. Information asymmetry is well known to be the main cause of market failure. How is that cabal alone better than that plus people with other specialties such as ethics, plus everyone able to watch them all and decide for themselves whether that aligns with their own (or society's) interests?
AI will change everything. Not in the distant future when it becomes sentient, but in the next 5-10 years.
ChatGPT may spit out word salad, and be trained on inappropriate input, but there is no doubting the success of some of these models, eg at playing the game of Go, predicting protein folding, etc, etc.
There's obviously something still missing, but the hype is warranted IMHO, these model represent a major advance compared to even just a few years ago.
There are good arguments in this article, buried in a mire of self righteous indignation.
Better education on viewing any output by a model with more critical eyes - which really is today a basic requirement for parsing any online information - is a fine bill to die on.
That then goes in hand with viewing with suspicious eyes the motivations of companies that are commercialising this technology, which is nowhere near a new sentiment.
However better recognition needs to be given to what obviously fuels the hype being railed against: the utility of this tool, even if it fails to work adequately most of the time. YMMV depending on your personal experience on the queried topics.
The article does indeed sound like it’s just preaching to the choir.