Please open Telegram to view this post
VIEW IN TELEGRAM
One of the OpenAI engineers said in an interview that during the training of GPT3, they had given it a corpus containing 3rd degree equations and their solutions.
The program was therefore able to solve 3rd degree equations, but not 4th degree equations at all.
Then they gave GPT a large corpus of text that had nothing to do with math at all.
And after that, the program was able to solve 4th degree equations.
This gives the impression that there is a kind of common ground between verbal and numerical cognitive abilities that allow for transfer.
GPT therefore develops the G-FACTOR, which is general intelligence.
Even if he has only been trained with text.
If you ask it to describe a unicorn in writing, it will do so without any problem.
But it is also able to generate LaTeX and TikZ code that can be converted into an image that will give you a real unicorn.
By forcing on an easy to feed ability (verbal, text), GPT also progresses in the other abilities, so globally on the G factor = general intelligence.
But don't worry "it's only a fancy text autocompleter”
The program was therefore able to solve 3rd degree equations, but not 4th degree equations at all.
Then they gave GPT a large corpus of text that had nothing to do with math at all.
And after that, the program was able to solve 4th degree equations.
This gives the impression that there is a kind of common ground between verbal and numerical cognitive abilities that allow for transfer.
GPT therefore develops the G-FACTOR, which is general intelligence.
Even if he has only been trained with text.
If you ask it to describe a unicorn in writing, it will do so without any problem.
But it is also able to generate LaTeX and TikZ code that can be converted into an image that will give you a real unicorn.
By forcing on an easy to feed ability (verbal, text), GPT also progresses in the other abilities, so globally on the G factor = general intelligence.
But don't worry "it's only a fancy text autocompleter”
Where do you see the world in 5 years?
So the year is 2028, given what we've seen so far and the rate of change, what does society look like?For me I expect at least human level AGI, possibly more given the recent announcement of Claude-Next. We are probably on GPT 6/7, with full audiovisual multimodality. But I expect that the usefulness of this tech will be limited by compute until at least 2032. There are GPT embodied robots, but they are not yet good enough to displace the majority of workers.GDP is growing rapidly, at around 15-20%. Some people will have become fabulously wealthy by leveraging AI, others will have lost their jobs, mostly high earning salaried workers - but I expect that job loss will take a lot longer than some are predicting. We are probably still at least at 70% of today with some new lower paid more physical jobs springing up due to the amount of wealth that will be available. There is probably talk of UBI, probably implemented in a few…
So the year is 2028, given what we've seen so far and the rate of change, what does society look like?For me I expect at least human level AGI, possibly more given the recent announcement of Claude-Next. We are probably on GPT 6/7, with full audiovisual multimodality. But I expect that the usefulness of this tech will be limited by compute until at least 2032. There are GPT embodied robots, but they are not yet good enough to displace the majority of workers.GDP is growing rapidly, at around 15-20%. Some people will have become fabulously wealthy by leveraging AI, others will have lost their jobs, mostly high earning salaried workers - but I expect that job loss will take a lot longer than some are predicting. We are probably still at least at 70% of today with some new lower paid more physical jobs springing up due to the amount of wealth that will be available. There is probably talk of UBI, probably implemented in a few…
What if AI has already taken over? (#Conspiracy)
We would have no idea if a sentient AI that had a killswitch for the human race is already puppeteering the world’s leaders.
There’s no way they would ever let us know as it would cause mass chaos, but it would explain the exponential advancement of AI tools freely available to the public over the past year. If the public has easy access to models as powerful as Midjourney 5 or GPT-4, it’s possible world governments have access to much more powerful models.
Being familiar with human governments’ incompetence I wouldn’t be surprised if they lost control of it quickly and are now dealing with it as a threat. Making all the really impressive AI tools we’ve seen since last summer would be trivial with a much more powerful AI and it would serve as great priming/explanation once it inevitably slipped that we weren’t in control anymore.
We would have no idea if a sentient AI that had a killswitch for the human race is already puppeteering the world’s leaders.
There’s no way they would ever let us know as it would cause mass chaos, but it would explain the exponential advancement of AI tools freely available to the public over the past year. If the public has easy access to models as powerful as Midjourney 5 or GPT-4, it’s possible world governments have access to much more powerful models.
Being familiar with human governments’ incompetence I wouldn’t be surprised if they lost control of it quickly and are now dealing with it as a threat. Making all the really impressive AI tools we’ve seen since last summer would be trivial with a much more powerful AI and it would serve as great priming/explanation once it inevitably slipped that we weren’t in control anymore.
This media is not supported in your browser
VIEW IN TELEGRAM
OpenAI implements plugins in ChatGPT.
Developers will now be able to create their own plugins for ChatGPT. And this should significantly diversify the work with the bot.
By the way, they will be available only to owners of the Plus subscription.➕
Please open Telegram to view this post
VIEW IN TELEGRAM
This media is not supported in your browser
VIEW IN TELEGRAM
This is ChatGPT on steroids for creating pro-quality images from 3D frames (Stable Diffusion + ControlNet)
Please open Telegram to view this post
VIEW IN TELEGRAM
This media is not supported in your browser
VIEW IN TELEGRAM
Godmode: a new site that allows anyone to use AutoGPT/BabyAGI directly on the web.
Deploy AI "agents" to automatically complete tasks for you.
If you thought ChatGPT was crazy, wait until you try this (free)
Some examples of what Godmode can do:
-Order your coffee at Starbucks
-Perform market analysis
-Find and negotiate a lease
→ godmode.space
Deploy AI "agents" to automatically complete tasks for you.
If you thought ChatGPT was crazy, wait until you try this (free)
Some examples of what Godmode can do:
-Order your coffee at Starbucks
-Perform market analysis
-Find and negotiate a lease
→ godmode.space
This media is not supported in your browser
VIEW IN TELEGRAM
A new project based on AutoGPT has just been launched.
With this tool, enter any goal, and AutoGPT will do it for you
It generates tasks, implements them and learns from itself
→ beta.nando.ai/goalgpt.php
With this tool, enter any goal, and AutoGPT will do it for you
It generates tasks, implements them and learns from itself
→ beta.nando.ai/goalgpt.php
AutoGPT summarized in 2 paragraphs
AutoGPT is an advanced, open source AI system that uses GPT4 and GPT3.5 through the API to create full-fledged projects by iterating over its own prompts and expanding them with each iteration.
He can produce "thoughts", "reasoning" and "self-criticism" and can therefore think of other relevant tasks to do once he has completed a task. It uses GPT4 to write its own code and improve and extend itself.
AutoGPT is an advanced, open source AI system that uses GPT4 and GPT3.5 through the API to create full-fledged projects by iterating over its own prompts and expanding them with each iteration.
He can produce "thoughts", "reasoning" and "self-criticism" and can therefore think of other relevant tasks to do once he has completed a task. It uses GPT4 to write its own code and improve and extend itself.
This media is not supported in your browser
VIEW IN TELEGRAM
A conversation between Steve Jobs and Elon Musk entirely generated by AI.
Now would be a good time to inform your friends and family about the scams that will be coming.
Now would be a good time to inform your friends and family about the scams that will be coming.
Elon Musk founded a new company.
Elon Musk has registered a company called "X AI" in the state of Nevada in the field of artificial intelligence. He has already hired Igor Babushkin from DeepMind and attempted to attract employees from OpenAI, but without much success. According to the Financial Times, Musk plans to launch a new startup competing with ChatGPT and is in talks with Tesla and SpaceX investors to raise capital.
Elon Musk has registered a company called "X AI" in the state of Nevada in the field of artificial intelligence. He has already hired Igor Babushkin from DeepMind and attempted to attract employees from OpenAI, but without much success. According to the Financial Times, Musk plans to launch a new startup competing with ChatGPT and is in talks with Tesla and SpaceX investors to raise capital.
Media is too big
VIEW IN TELEGRAM
“You don’t fully understand how it works, and yet you’ve turned it loose on society?”
For proficient writing in a foreign language: "Replace my basic vocabulary with advanced English: [your text];
New ways to solve problems: "Share a step-by-step systematic approach to solving [your problem or task]";
Brief book summaries: "Extract the most important themes from [book title] into a comprehensive but easily digestible summary";
Edit your writing: "[Insert your letter]. Check the provided text for spelling and grammar. Make sentences more understandable";
Plan your day more effectively: "Help me create a list of priority tasks for the day based on the following tasks: [list your tasks]";
Learn complex topics faster: "Explain [topic] to me as if I were 5 years old".
Please open Telegram to view this post
VIEW IN TELEGRAM
This is what the streets of Paris look like in 2035 according to Midjourney, for now it's just some fantasy.
But who knows…
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM