Corn Yields & AI Model Sizes
Strangely similar charts, right?
Turns out they’re even more similar than you’d think.
(1) Pre-existing tech paths that industries refused to sufficiently invest in: Both a case of a long-time refusal to make investment, in a new path, despite that it was already well-established for many years that that new path would produce the outsized gains, if only sufficient investment was made, and yet they continued to refuse.
(2) But once they did, no going back: Both cases where once anyone in the industries finally switched to making the big investments, they entered a new era of ongoing improvement that continued indefinitely.
Sometimes the answer is already there, and big investment is all you need, but industries will collectively just refuse to do it, for years.
Usually with the excuse that there must be some more special secret sauce, that some stronger “social proof” is needed before bigger investment.
Brother in AI, the recipe is public.
The secret sauce is sufficient money, invested in that new path.
Bitter Lesson once again.
Strangely similar charts, right?
Turns out they’re even more similar than you’d think.
(1) Pre-existing tech paths that industries refused to sufficiently invest in: Both a case of a long-time refusal to make investment, in a new path, despite that it was already well-established for many years that that new path would produce the outsized gains, if only sufficient investment was made, and yet they continued to refuse.
(2) But once they did, no going back: Both cases where once anyone in the industries finally switched to making the big investments, they entered a new era of ongoing improvement that continued indefinitely.
Sometimes the answer is already there, and big investment is all you need, but industries will collectively just refuse to do it, for years.
Usually with the excuse that there must be some more special secret sauce, that some stronger “social proof” is needed before bigger investment.
Brother in AI, the recipe is public.
The secret sauce is sufficient money, invested in that new path.
Bitter Lesson once again.
🔥10💯5😢3
Media is too big
VIEW IN TELEGRAM
Professor and associate editor at a journal makes the case for why it’s ok to use ChatGPT to write scientific papers
👍11🤣4
Number of papers published per month in the arXiv AI categories has been growing exponentially for decades
Nature Article
Nature Article
🔥13🤣3
Forwarded from Chat GPT
Fairwell text-davinci-003, i.e. the original, powerful GPT-3.5
Soon, January 6th, the last remaning nearly-pure powerful foundation model, not contaminated by OpenAI's moralizing, gets unplugged
Future is filled with foundation models pushing a certain set of values on you.
Soon, January 6th, the last remaning nearly-pure powerful foundation model, not contaminated by OpenAI's moralizing, gets unplugged
Future is filled with foundation models pushing a certain set of values on you.
😢17🤬6🤣2🫡2
Intelligent brains take longer to solve difficult problems
“Brains with reduced synchrony between brain areas literally “jump to conclusions” when making decisions, rather than waiting until upstream brain regions could complete the processing steps needed to solve the problem”
“Synchronization, i.e., the formation of functional networks in the brain, alters the properties of working memory and thus the ability to ‘endure’ prolonged periods without a decision”
Are you able to endure prolonged periods of thinking without jumping to a conclusion?
Article
“Brains with reduced synchrony between brain areas literally “jump to conclusions” when making decisions, rather than waiting until upstream brain regions could complete the processing steps needed to solve the problem”
“Synchronization, i.e., the formation of functional networks in the brain, alters the properties of working memory and thus the ability to ‘endure’ prolonged periods without a decision”
Are you able to endure prolonged periods of thinking without jumping to a conclusion?
Article
✍17🤣6🔥4🤯2👀2💯1
Media is too big
VIEW IN TELEGRAM
Year of the AI Girlfriend is Almost Here
“Introducing 𝐌𝐨𝐛𝐢𝐥𝐞 𝐀𝐋𝐎𝐇𝐀🏄 A low-cost, open-source, mobile manipulator.”
“In this work, we develop a system for imitating mobile manipulation tasks that are bimanual and require whole-body control. We first present Mobile ALOHA, a low-cost and whole-body teleoperation system for data collection. It augments the ALOHA system with a mobile base, and a whole-body teleoperation interface. Using data collected with Mobile ALOHA, we then perform supervised behavior cloning and find that co-training with existing static ALOHA datasets boosts performance on mobile manipulation tasks. With 50 demonstrations for each task, co-training can increase success rates by up to 90%, allowing Mobile ALOHA to autonomously complete complex mobile manipulation tasks such as sauteing and serving a piece of shrimp, opening a two-door wall cabinet to store heavy cooking pots, calling and entering an elevator, and lightly rinsing a used pan using a kitchen faucet.”
Tutorial
Github
Project Website
“Introducing 𝐌𝐨𝐛𝐢𝐥𝐞 𝐀𝐋𝐎𝐇𝐀🏄 A low-cost, open-source, mobile manipulator.”
“In this work, we develop a system for imitating mobile manipulation tasks that are bimanual and require whole-body control. We first present Mobile ALOHA, a low-cost and whole-body teleoperation system for data collection. It augments the ALOHA system with a mobile base, and a whole-body teleoperation interface. Using data collected with Mobile ALOHA, we then perform supervised behavior cloning and find that co-training with existing static ALOHA datasets boosts performance on mobile manipulation tasks. With 50 demonstrations for each task, co-training can increase success rates by up to 90%, allowing Mobile ALOHA to autonomously complete complex mobile manipulation tasks such as sauteing and serving a piece of shrimp, opening a two-door wall cabinet to store heavy cooking pots, calling and entering an elevator, and lightly rinsing a used pan using a kitchen faucet.”
Tutorial
Github
Project Website
🔥10🥰3🤣3👏1😇1
This media is not supported in your browser
VIEW IN TELEGRAM
It only costs $7k
“Goals:
1. Moves fast. Similar to human walking of 1.42m/s.
2. Stable. Manipulate heavy pots, a vacuum, etc.
3. Whole-body. All dofs teleoperated simultaneously.
4. Untethered. Onboard power and compute.”
“To achieve these goals, we mount ALOHA to a mobile base designed for warehouses: Tracer AGV”
“It can carry 100kg, move up to 1.6m/s, while costing only $7k”
Mobile ALOHA
“Goals:
1. Moves fast. Similar to human walking of 1.42m/s.
2. Stable. Manipulate heavy pots, a vacuum, etc.
3. Whole-body. All dofs teleoperated simultaneously.
4. Untethered. Onboard power and compute.”
“To achieve these goals, we mount ALOHA to a mobile base designed for warehouses: Tracer AGV”
“It can carry 100kg, move up to 1.6m/s, while costing only $7k”
Mobile ALOHA
🔥24👀8🤣1
Make race-based and sex-based hiring illegal again.
Overturn Griggs v. Duke Power Co.
Eliminate Disparate Impact.
Competence Over Quotas.
Their long march has only just begun.
Join the fight: @chadgptcoin
Tweet Link
Overturn Griggs v. Duke Power Co.
Eliminate Disparate Impact.
Competence Over Quotas.
Their long march has only just begun.
Join the fight: @chadgptcoin
Tweet Link
🤬26🗿7👍6💯2😘2🤣1
The AI company’s social media manager endorsed this guy who says what they’re doing is wrong.
Why?
They agree that it’s wrong.
Why post job ads like this then?
It’s the law.
Boobs over brains, less white or goodnight.
The US law has what are referred to as legal “tensions”, or more appropriately, outright Maoist contradictions.
So then, how is a company to decide which out of two contradictory laws they must follow?
Zeitgeist of the time.
Back in 2009, the slide toward forced-discrimination by race and sex was already well underway, but not yet complete, with the anti-discrimination side still experiencing some wins.
Today?
The zeitgeist has shifted.
Today companies absolutely MUST discriminate by sex and race, or else.
The AI company here is legally right, and anyone with a genuine sense of right and wrong is upset about it.
Boobs over brains, less white or goodnight.
Disparate Impact Legal Tension
Why?
They agree that it’s wrong.
Why post job ads like this then?
It’s the law.
Boobs over brains, less white or goodnight.
The US law has what are referred to as legal “tensions”, or more appropriately, outright Maoist contradictions.
So then, how is a company to decide which out of two contradictory laws they must follow?
Zeitgeist of the time.
Back in 2009, the slide toward forced-discrimination by race and sex was already well underway, but not yet complete, with the anti-discrimination side still experiencing some wins.
Today?
The zeitgeist has shifted.
Today companies absolutely MUST discriminate by sex and race, or else.
The AI company here is legally right, and anyone with a genuine sense of right and wrong is upset about it.
Boobs over brains, less white or goodnight.
Disparate Impact Legal Tension
💯8🤬4🥰2🤣1
AI Safety
“The west has shifted away from a culture of progress in recent decades, and towards one of caution, worry and risk-aversion, with economic growth slowing”
Is the west talking itself into decline?
“The west has shifted away from a culture of progress in recent decades, and towards one of caution, worry and risk-aversion, with economic growth slowing”
Is the west talking itself into decline?
🤬18👍10🥰1🤣1🗿1