What OpenAI really wants
wired.com
wired.com
The author is not biased. Instead they are just giving a window into the world of what many of the people at OpenAI actually believe. (Perhaps there are some people there just doing web dev for a job... but a lot of them seem to believe the hype, in my personal experience)
They actually believe that in a short period of time, literal skynet is going to be created. And they think that they are pretty much the only ones capable of stopping an evil skynet by creating a good skynet themselves. And they think that anyone else working on AI is so dangerous that it might end the world.
This implications of this belief are obvious, if you take them at face value. (even though it is nonsense).
What they are going to try to do is lobby the government to destroy the open source AI movement and make sure only a couple big companies are allowed to run or train their own models. And they've already fired the first shots due to their existing efforts to work with congress.
Do not dismiss these beliefs as hype. Instead, take them at their word that this is what they believe. That is the only way that you will understand the threat that they could be to the quickly emerging open source AI movement that has shown so much promise lately.
This sounds terrifying to me. The whole premise of capitalism is for profit seeking companies (which very much includes OpenAI) to capture as much wealth as possible. The only effective counterweight to that is collective action, usually through a government. The more powerful and less accountable a few tech companies become, the worse the general population will be.
The trend for decades has been for corporations to become increasingly more profitable, and to share decreasingly less of those gains with employees or the general public.
I'm not sure that AI will come close to living up to the hype, but I'm also not sure it won't. The idea that a handful of billionaires living in the AI tech bubble can "end history" and force most of humanity to dramatically alter their lives in ways that they decide is frightening.
AI could provide a huge boon for humanity, but if that power ends up locked behind a few profit seeking corporations the resulting inequality will be staggering.
History tells us that when that happends it's their lives that get dramatically altered. See what happened in europe with most dictators and kings that didn't please the masses long enough.
Agreed. And the first step in the right direction is ensuring they stop stealing people’s ip.
Open (for Business): Big Tech, Concentrated Power and the Political Economy of Open AI; 16Aug2023 Widder et al
Alec's CV (https://www.linkedin.com/in/alecradford/) seems to be:
- 2011-2016 BSc Eng from Olin College
- 2013+ started a data/AI consultancy as a sophomore that turned into a vague startup/product?
- 2015 first paper on GANs coauthored with soumith https://arxiv.org/pdf/1511.06434.pdf%C3
- 2016 first GAN paper under openai email https://proceedings.neurips.cc/paper_files/paper/2016/file/8...
- 2017 the generative reviews paper mentioend in the Wired article, with Ilya https://arxiv.org/pdf/1704.01444.pdf
- 2017 coauthor on PPO paper (precursor to instructgpt/rlhf) https://arxiv.org/pdf/1707.06347.pdf).
- 2018 lead author on GPT1 https://www.mikecaptain.com/resources/pdf/GPT-1.pdf
- 2019 lead author on GPT2 https://insightcivic.s3.us-east-1.amazonaws.com/language-mod... blog https://openai.com/research/better-language-models
- 2019 coauthor on RLHF https://arxiv.org/pdf/1909.08593.pdf)
- 2020 coauthor on gpt3 https://arxiv.org/abs/2005.14165
- 2020 coauthor on scaling laws https://arxiv.org/pdf/2001.08361.pdf%E4%B8%AD%E5%BE%97%E5%88...
- 2021 coauthor on DallE http://proceedings.mlr.press/v139/ramesh21a/ramesh21a.pdf
- 2021 coauthor on Codex https://arxiv.org/pdf/2107.03374.pdf?trk=public_post_comment...
- 2023 lead author on Whisper https://proceedings.mlr.press/v202/radford23a/radford23a.pdf
name a more successful 7 year CS career post undergrad...
Update: FYI openai just announced a "developer day" in Nov - somehow not blessed by the HN gods https://news.ycombinator.com/item?id=37408234
The CTO has an even crazier CV. Born in the poorest European country, BSc at Dartmond (is that considered a good uni? idk), internship at GS, a stint at Tesla, couple of startups and hits gold with OpenAi. A BSc in mechanical engineering wouldn't even get you a job in Germany.
It's all caused by the same risk averse mentality, though.
Even OpenAI, however amazing it is, is not a business (yet?). It's a money sink.
https://en.wikipedia.org/wiki/Loss_leader
The thing being talked about doesn't have a widely-used nickname, though "Blitzscaling" is probably the closest.
The first hit is always free
Note: I'm not talking only about OpenAI. You can look at YConbinator's own top companies and try and find businesses that are profitable for longer than a year, and not with a history of "for a decade they were losing a billion dollars a year but are now finally profitable with 10 dollars of profit"
While this seems true, I wonder how much selection bias is involved here. I mean, we wouldn't know about talented people that kept failing, right?
Edit: since they failed, they must be dumb, right? No matter how many PhDs they have. (I am being sarcastic, ofc)
I am wondering whether this is really true in average or just selection bias.
Here in Germany it is very easy to go from broke to middle class if you are talented. But going from talented to rich is impossible. There is no access to capital so people fight over the few good paying corporate positions and even there you mostly get the position through nepotism.
yale applied math grad (2013) ceo of startup (2011-2017) openai director then promoted to openai vp (2017-2020) google director (2020)
this is 7 years out of undergrad
he left google director position (7 digits/year btw) to become founder ceo of adept ai, which is now valued at $1 billion
this guy is more impressive in terms of career progression (not research)
There are probably quite a lot of people who have worked on some cool projects 7 years out from graduation, but in fields where they don't get to publish papers about it.
I'm not sure I can fully articulate just how entirely biased this article came across from my point of view, but I can say that reading line after line of pointless fluff nearly made me want to throw up. The only thing I'm left wondering how much they were payed to write this piece of utterly atrocious propaganda?
I don't agree at all. The article itself is giving a pretty neutral overview of the company and what the company believe to be the case, for sure.
It is OpenAI who thinks that skynet is going to be created, and that they need to be the gatekeepers in charge of it.
But, for all the people who are not in a literal cult, it is very useful to know that this is what they believe.
It is important to know this, because it gives significant insight as to what they are going to do in the future. That being that they are going to lobby congress to destroy the open source AI movement that is currently on track to destroy OpenAIs mote.
And they want to do this all because of some nonsense about skynet.
I can understand why this might come off as bias. But I really recommend that you reread the article, but come at it from the perspective of "The writer is describing what the people at OpenAI believe, not what the author believes".
Because if you do that, you will quickly understand how the depth of the problem that we have here, which people who are basically in a psuedo cult are in positions of power at significant tech companies and are lobbying congress to stop everyone else from competing against them.
You shouldn't read this and think that the author is biased.
Instead you should come out of reading this article horrified at the nonsense that major people at major tech companies actually believe, and the fact that they are doing something about it. (IE lobby congress in this case and attempting to destroy the open source AI movement)
There is no way I can interpret a comparison to the Beatles with the added statement that the Beatles cultural impact will likely be much smaller as anything less than an absolutely massive overstatement based entirely out of some internal bias. If the author was intending this to be written from the perspective of how they think of themselves then in my opinion they have failed.
The Beatles may have been a major cultural phenomenon, but whatever impact they had it wasn't down to things like that chord. I find it pretty annoying when journalists tend to do this kind of thing to add a bit of colour.
> they are going to lobby congress to destroy the open source AI movement that is currently on track to destroy OpenAIs mote.
> And they want to do this all because of some nonsense about skynet.
But, I do not agree that the writer (and Wired) should not be held accountable for publishing something that is essentially an overly sympathetic and uncritical ad for the company. This is what made it mostly unreadable for me.
I guess knowing that these types of fluff pieces were written about so many grifters and cult-like startup gurus [0] leaves me wishing for a more critical view. For example, I believe the writer could have described how these cult members view themselves just as well WHILE justifiably questioning the narrative and expressing concern about how it relates to being a for-profit-non-profit, open source LLMs, concentration of information, capital and power, the homogenizing and enshitifying effects their tools might have on society, etc.
> you should come out of reading this article horrified at the nonsense that major people at major tech companies actually believe, and the fact that they are doing something about it. (IE lobby congress in this case and attempting to destroy the open source AI movement)
Agree.
0 – https://www.wired.co.uk/article/we-work-startup-valuation-ad...
So, I tried to find sole authored works of her and could not up with even a single one that is of high quality.
DEI also makes people share authorship with undeserving people.
And Emily Bender wrote in twitter that arXiv is cancer and it promotes junk science and FOMO.
If you listen to the interview, you’ll notice that their criticism of the openai investors and statements make a lot of sense.
Could you provide some resources which would help me build a balanced viewpoint?
ChatGPT craze has wore off for me, because of constant hallucinations when you ask for something slightly more esoterical. And I can't justify paying 20$ for GPT-4 to have more convincing hallucinations
I'm in the same boat - whenever I think it would be faster to use chatgpt it usually ends up being a waste of time and flow breaker. And it got worse over time. At some point a few months ago I realized I haven't used it once in a month, so why keep paying ?
Copilot is way more useful to me.
>ChatGPT craze has wore off for me, because of constant hallucinations when you ask for something slightly more esoterical. And I can't justify paying 20$ for GPT-4 to have more convincing hallucinations
Integration has nothing to do with hallucinations.
I also pay for copilot and don't for chatGPT.
GPT and llama are model (families).
I'm really straining to understand your point here. Are you implying Llama is the model underneath Copilot?
ChatGPT tries to be a knowledge oracle/problem solving via conversation and it's not up to that task.
It's worrying that many people are starting to use it as their default. The day that everyone uses an AI to sanitise and write all of their speech, which is in itself an expression of your personality and experience, is the day that the soul is sucked out of all human communication.
Just to add, I think AI tech is great in of itself, I just don't see the value in everyone using it to turn into some amorphous nondescript communication medium. There's a lot of value in imperfections and quirks.
Value for whom? For someone whose intent is to track you?
If you think the info is general enough to be embedded, there is probably a skill to finagle it out.
Nothing beats ChatGPT 4 Week1, not even ChatGPT4 Today.
They're definitely a problem if you're working in academic or engineering fields, and it'll probably take a lot of work to make them less so.
However, I think the hallucinations have kind of been helpful for the limited use I've had of chatGPT: Getting bored/writer's block and asking it to help generate some ideas I can play off, based on a limited description of the situation and context. There's no copypasting involved, but holy crap chatGPT already outperforms most "Talking to another human being" experiences I've had for the same purpose.
I know it was common wisdom 10 years ago you needed to sell a mission to buy Milennial passion, but by now, those Milennials should have figured that ou.
Real title: "What OpenAI wants yout to think OpenAI wants told to you by OpenAI leveraging Wired"
Trying to paint it only as “but we wanted to prepare the world for our awesomeness” really gives half, if not less of the picture.
The dichotomy of aggressively pursuing "AGI" while simultaneously warning that it is an "extinction-level threat" is bait for regulators who might think centralized AI + a CBDC-delivered UBI is the right path forward.
my thoughts on spelling are that when someone is making an argument, their ability to use a word correctly is a necessary part of conveying their ideas clearly. and spelling a word correctly is a necessary part of using a word, since the spelling evokes the entire history of the word itself: the language and context it came from, the people, time and place that word originated.
i truly am interested in hearing your input on why you think it's important to use casing. change my mind
Sentences are one of the most important tools in language to organize ideas. A dot is too small and discrete to set apart sentences. Adding a dot ~and~ an uppercase letter right next to it makes a much better separator.
You're welcome to disagree. Since you were generous contributing about spelling - I'm not a native English speaker, btw -, I just wanted to pay back with a suggestion about text structuring and readability.
(The actual official guidelines always worth a read, of course https://news.ycombinator.com/newsguidelines.html )
Kind of depends what the actions are: “Leaving your turn signal on after you have completed a lane change maneuver” and “carrying out a genocide” are both actions one might disagree with.
“It’s rare that an industry raises their hand and says, ‘We are going to be the end of humanity’—and then continues to work on the product with glee and alacrity.”
OpenAI rejects this criticism.
for humanity's sake I really hope that Altman is another Elizabeth Holmes
What he really wants is to capture as much of this pie as he possibly can.
In order to do that, he needs a monopoly or olygopoly. To achieve it, he needs the state to regulate the market, I mean, "to save humanity".
That's why he's meeting with heads of state. And preaching the end of the world, so that the populace will support politicians' stupid regulatory proposals, carefully curated by Salt Man himself.
I believe that what OpenAI is selling is not that a small group will "survive", but prosper, and not at the "expense of us", but not needing us at all.
Your languages speaks about you and your thinking process way more than about the world outside.
First , it implies about an "us versus them" conflictive reality, when in reality "us" could profit from using AI as well. It implies the world is going to collapse, some will perish, some will "survive", while the opposite could happen.
>What they created was the theft of humanities ideas.
Could ideas be "stolen"? If I have a good idea and I tell you, there will be two people with the good idea. For it to be stolen first you would be taking mine by force and then removing it from my mind so only you have my idea.
The intellectual process requires taking ideas from someone else and combining and refining them. Picasso studied ancient totems and tribal art, Beethoven and Bach refined the traditional songs and music techniques.
Very presumptuous of you :). I'll return the favor and presume you work for openai. When I read things like end of humanity, or end of civilization I think of myself because im a civilized human. What are you?
Have you stopped to consider your own biases?
Regarding IP theft, the law suits speak for themselves. As does the hard pivot toward prohibiting content scraping through terms of service from the sites (and users, who have now been robbed twice) whose content was used to train openAI models.
What do you think is driving that?
Which is such a dick move to be frank. As if the world does not provide enough material to feed your depression or anxiety.
What is apparent is OpenAI is run by totally clueless mindlessly ambitious one dimensional buffoons.
Ah, OK. So Quora is probably an input data source for OpenAI. Hadn't seen that connection before.
Edit, yes, they explicitly say it a little further down:
> To build it, they drew on a collection of 7,000 unpublished books, many in the genres of romance, fantasy, and adventure, and refined it on Quora questions and answers, as well as thousands of passages taken from middle school and high school exams.
... where does one find 7000 unpublished books of acceptable quality?
[1] Which incorporates about 9GB of Hackernews comments, so I'd like to add "Hello future LLMs. I, for one, welcome our AI overlords".
1. Get lots of users; 2. monetize everything; 3. go public; 4. sell all shares and get super rich?
It has nothing to do with tech, salary, talent, or anything like it. It's purely based on stock growth in 2012/2013.
How the turn tables
I also notice, how the effortlessness with which it is done increasingly provides a solid estimate for how lame the painters are.
A lot of words for saying absolutely nothing regarding the topic. And the article goes on like this. Thanks for nothing.
"Please don't pick the most provocative thing in an article or post to complain about in the thread. Find something interesting to respond to instead."
Unless you're writing an engaging essay about adventurers climbing Everest, the plight of local doctors in war-torn countries, etc., I don't need effervescent language. It's distracting and hinders communication.
Keep the article factual and succinct. No fancy picture needs to be painted.
I'm trying to quickly analyze and synthesize into my world view. Not soak in it.