A doozie @GaryMarcus

Hear, hear. And bonus - it doesn’t sound like there are plans to embed annoying AI assistants in Linux

‘THEY’RE EATING THE DOGS’ - Trump, A Dramatic Monologue

In my entire career, I have never seen an experiment fail so many times, at such great cost, as the experiment on the “scaling” hypothesis that you could create AGI simply by adding more data and compute. It doesn’t work! Try something different, already?

𝙒𝙝𝙖𝙩 𝙝𝙖𝙥𝙥𝙚𝙣𝙨 𝙞𝙛 𝙩𝙝𝙚 𝙂𝙚𝙣𝘼𝙄 𝙗𝙪𝙗𝙗𝙡𝙚 𝙘𝙤𝙡𝙡𝙖𝙥𝙨𝙚𝙨? [Sequel to tweet below that 1 million people viewed] • AI startups will face 𝗮 𝗺𝘂𝗰𝗵 𝘁𝗼𝘂𝗴𝗵𝗲𝗿 𝗶𝗻𝘃𝗲𝘀𝘁𝗺𝗲𝗻𝘁 𝗰𝗹𝗶𝗺𝗮𝘁𝗲. Many will fold, others will be acquired. • Investors will 𝗹𝗼𝘀𝗲 𝗮 𝗹𝗼𝘁 𝗼𝗳 𝗺𝗼𝗻𝗲𝘆. Some pension funds will take a hit. • 𝗠𝗦𝗙𝗧 𝘄𝗶𝗹𝗹 𝗱𝗿𝗼𝗽. Perhaps NVidia as well, as chip demand declines. • 𝗔𝗜 𝘄𝗼𝗻’𝘁 𝗱𝗶𝘀𝗮𝗽𝗽𝗲𝗮𝗿. You will still have GPS in your car, Google Search, and a bunch of chatbots you can play with. Meta will keep giving trained LLMs away for free. • Coders will still use it, but 𝘁𝗵𝗲 𝗽𝘂𝗯𝗹𝗶𝗰 𝘄𝗶𝗹𝗹 𝗹𝗼𝘀𝗲 𝗶𝗻𝘁𝗲𝗿𝗲𝘀𝘁, 𝗮𝗻𝗱 𝘀𝗼 𝘄𝗶𝗹𝗹 𝗺𝗮𝗻𝘆 𝗰𝗼𝗿𝗽𝗼𝗿𝗮𝘁𝗲 𝗰𝗹𝗶𝗲𝗻𝘁𝘀. • Grifters will 𝗺𝗼𝘃𝗲 𝗼𝗻 𝘁𝗼 𝗼𝘁𝗵𝗲𝗿 𝘁𝗵𝗶𝗻𝗴𝘀. • You will 𝘀𝘁𝗼𝗽 𝗵𝗲𝗮𝗿𝗶𝗻𝗴 𝗮𝗯𝗼𝘂𝘁 𝗔𝗜 𝗲𝘃𝗲𝗿𝘆 𝘀𝗶𝗻𝗴𝗹𝗲 𝗱𝗮𝘆. • 𝗕𝗮𝗱 𝗮𝗰𝘁𝗼𝗿𝘀 𝘄𝗶𝗹𝗹 𝗸𝗲𝗲𝗽 𝘂𝘀𝗶𝗻𝗴 𝗟𝗟𝗠𝘀 𝘁𝗼 𝗰𝗿𝗲𝗮𝘁𝗲 𝗺𝗶𝘀𝗶𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝗽𝗵𝗶𝘀𝗵 𝗰𝗿𝗲𝗱𝗲𝗻𝘁𝗶𝗮𝗹𝘀. 𝗗𝗲𝗲𝗽𝗳𝗮𝗸𝗲𝘀 𝗮𝗿𝗲𝗻’𝘁 𝗴𝗼𝗶𝗻𝗴 𝗮𝗻𝘆𝘄𝗵𝗲𝗿𝗲. The need for AI regulation will remain urgent. • The field of AI research will go on, but 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 𝘄𝗶𝗹𝗹 𝗺𝗼𝘃𝗲 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗳𝗼𝗿𝗲𝗴𝗿𝗼𝘂𝗻𝗱 𝘁𝗼 𝘁𝗵𝗲 𝗯𝗮𝗰𝗸𝗴𝗿𝗼𝘂𝗻𝗱, as better approaches are eventually invented. • AGI will eventually come, but certainly 𝗻𝗼𝘁 𝘁𝗵𝗶𝘀 𝗱𝗲𝗰𝗮𝗱𝗲.

A former political prisoner on how the West should honour Alexei Navalny’s legacy economist.com/by-invitation/… from @TheEconomist

A question for #pyschologists and #behaviouvalScientists - I figure that the more you know about something, the more complex your behaviour in that domain. Does that assumption ignore early stage exploration? Do we become predictable as we develop knowledge? Upside down U?

You know life is hectic when cleaning your bathroom counts as #selfcare

Exactly this. Props

If AI apps don’t improve productivity, they won’t be economically viable, no matter how fascinating they are to play with

It’s Saturday morning and I have stuff to do. But it’s all a bit daunting. So I’ll make a list of items to knock down. Am I #procrastinating or am I being efficient - or is procrastination efficient sometimes?

Republicans Accuse Democrats of Plotting to Nominate Winning Candidate borowitzreport.com/p/republi…

Karl Friston says that current AI language models lack an essential quality of intelligent agents and that is curiosity: "to be an agent is to be curious"

Is a 'global technical outage' an argument for #opensource software? I'm guessing it is. Anyway, I should get back to work #ILoveLinux

LLMs went from revolutionary to “charming”, from age 30 to age 80 in like 2 years.

Patio, pizza and the incredible ⁦@carlorovelli⁩ . My kind of urban holiday

Important reading for anyone serious about #AGI cell.com/trends/cognitive-sc…

Francois Chollet: OpenAI has set back the progress towards AGI by 5-10 years because frontier research is no longer being published and LLMs are an offramp on the path to AGI

You can read some email exchanges between OpenAI and ex-employees over at vox.com/future-perfect/35113…. There are a lot of forms of courage, but this sure is one of them.