🔥Telegram İfşa

Re the looney billionaire who thinks he’s anti-aging. All I’ll say is … 1) weak sauce science. 2) so… tell us more about your “brain Fitbit” (lol at the idea you can monitor the brain non-invasively via photons sent from outside the skull).

Crypto red flag: using Telegram. theguardian.com/world/2023/feb…

The problem isn’t AGI-alignment*, it’s human-alignment, because humans are the ones who wield AI. *we can’t extrapolate from text completion to generalized intelligence, when there does not exist a testable hypothesis of consciousness.

Please someone make an LLM and call it WinAmp.

Soon a North Korean LLaMA will be scamming folks... influencing folks... running an AI-"romantic partner" app. Exciting times.

It's a good thing deposits at SVB and Signature are backstopped. It's misleading to call it a bailout, and ... similarly misleading to say that the taxpayer bears no cost when FDIC premiums are a cost to banks, and obviously banks pass costs on to their (tax-paying) customers.

FTX investor.

GPT-4 is *impressive*, but “Open” AI, it ain’t. It’s a good thing that poking it via APIs will let folks distill more performant models (eg for phones) from it. Still concerning that it’s internal state is utterly inscrutable (moreso after distilling?).

What happens to copyright when 1. Someone trains a model on the top 10,000 films and TV shows 2. Anonymously releases the weights via BitTorrent And 3. Folks distill this quite-inscrutable network into smaller ones?

The chaos in France will come to Canada within a decade if we can’t prioritize meaningful investment in innovation. Not what Chrystia Freeland talks about… but putting billions into nuclear power, biological computing, secure computing, and offensive cyber weapons, to start.

Does she misunderstand the implications of binary apps on phones and the Chinese gov’t legal authority to compel its companies to do as they choose? Or is she cynically trying to gain votes from uninformed TikTok voters? Sigh, I truly was a fan, too.

The reason TikTok and other apps with primarily Chinese development teams (like Zoom) ought to be restricted in the US. The west backdoored tonnes of infrastructure sold to foreign nations. Why wouldn’t they do the same? ⁦Alexandria Ocasio-Cortez⁩ arstechnica.com/information-te…

When UofTs shame is your poster-boy…

What if the biggest consequence of (open-source) AI is that it makes each human infinitely more interesting, aware, creative and productive; and so the cost of coercing any human asymptotes to infinity?

Is testosterone safer (better, cheaper) for me than Ozempic? open.substack.com/pub/suthakamal…

Hi James Cham - here is the first video clip (3min) from that session - 𝐒𝐞𝐧𝐝𝐡𝐢𝐥 𝐌𝐮𝐥𝐥𝐚𝐢𝐧𝐚𝐭𝐡𝐚𝐧 and Sutha Kamal - exploring the ability of AIs to impart a form of scientific understanding that each of us yearns - an understanding of ourselves Creative Destruction Lab (CDL) tinyurl.com/yd55pxcc