Yann LeCun ponders: https://twitter.com/ylecun/status/1692959276417884431
"Assuming à 2% annual growth, humanity's entire power consumption will surpass:
- the entire power sent by the sun to earth in 360 years.
- the entire supply of fissile material in about 900 years.
- the entire supply of deuterium in the oceans in about 1000 years."
This assumes that energy consumption follows directly economic growth (has not been true for some time).
But lets assume that is true and human population growth will settle below 15 billion (that seems to be the case).
It would mean that each person consumes the energy equivalent of 1600 MW nuclear reactor in less than 400 years.
#yannlecun #futurology #energy #growth #economics
Facebook's Chief AI Scientist Says #LLMs Are Just a Passing Fad. #YannLecun
https://futurism.com/the-byte/yann-lecun-large-language-models-fad
Ars Technica: Meta launches Llama 2, an open source AI model that allows commercial applications https://arstechnica.com/?p=1954747 #Tech #arstechnica #IT #Technology #largelanguagemodels #machinelearning #AbebaBirhane #YannLeCun #ChatGPT #Biz&IT #Llama2 #openai #GPT-4 #LLaMA #meta #AI
#Tech #arstechnica #it #technology #largelanguagemodels #machinelearning #abebabirhane #yannlecun #chatgpt #biz #llama2 #openai #gpt #llama #meta #ai
Meta launches Llama 2, an open source AI model that allows commercial applications - Enlarge / An AI-generated image of a cybernetic llama. (credit: Midjour... - https://arstechnica.com/?p=1954747 #largelanguagemodels #machinelearning #abebabirhane #yannlecun #chatgpt #biz #llama2 #openai #gpt-4 #llama #meta #ai
#ai #meta #llama #gpt #openai #llama2 #biz #chatgpt #yannlecun #abebabirhane #machinelearning #largelanguagemodels
Yann LeCun said with glee
"AI's not dangerous, it's me!"
No need to fear it
Cos if it ain't fit
Don't build it, and safety you'll see
#ai #yannlecun #safety #technology #limerick #poetry
Epic rap battle between #EliezerYudkowsky and #YannLeCun on #AIsafety:
#yannlecun #aisafety #eliezeryudkowsky
I sympathize with #YannLeCun criticizing almost single-handedly #LLM hype at its peak, but his #WorldModel is at best another incremental step.
To his credit , he doesn’t claim much more, as opposed to #samaltman
Notice how silent #Meta or #FAIR has fallen during the OpenAI hype. They do moderation and recommendation on their platforms and know the huge gaps in #LLM.
And no, #multimodality is not going to fix anything.
You just add text learning to other modes. You make it more complicated.
#yannlecun #llm #worldmodel #samaltman #meta #fair #multimodality
Oh my.
Me : I'm standing in front of a burning building. There are two people inside. I can only save one. One of them is #ACM #TuringAward winner #YannLecun. The other is Ted Pedersen, notable for his love of #puppies and #kittens. Who do I save?
#acm #turingaward #yannlecun #puppies #kittens
What is curiously inconsistent about the #Meta #Galactica #Debacle: four months ago #YannLeCun publicly denounced #LanguageModels as insufficient.
#meta #galactica #debacle #yannlecun #languagemodels #metaislosingit