Paul Cook (@langabi) profile picture

https://nitter.cf/t.co/5KHQcT2MBp Opinions are my own

Johannesburg
Joined April 2008
Photo by Paul Cook (@langabi)Photo by Paul Cook (@langabi)Photo by Paul Cook (@langabi)Photo by Paul Cook (@langabi)Photo by Paul Cook (@langabi)Photo by Paul Cook (@langabi)
In 2019, humans mined+quarried more by weight than in all history from the start until 1950. Crazy! [Material World by Ed Conway, lnk.to/MaterialWorld]
34
Today, in what NOT to do with Whatsapp support, from @supersonicfibre : - no response for 18 hours - then useless bot that can't detect slow response (why only now? Why not fix bot?) - then actually Whatsapp support not working at all - then 26 mins trying to phone, no reply
1
2
234
Freeman Dyson's detailed analysis of how life can survive through thermodynamic optimization in an expanding universe: this paper literally defines the words "audacious" and "creative", and it's not just speculation - Dyson brings hard numbers to the table panspermia.org/revmodphys.51…
3
36
1
455
75,123
Paul Cook retweeted
most people don’t realize a major part of being a corporate executive is slack jiu jitsu. it’s literally just maintaining state on a million ongoing threads and connecting person A to person B to resolve dependency conflicts at an incredibly rapid pace
82
129
40
2,625
338,818
Everything that has happened, must have been inevitable. [Slight exception for quantum-scale phenomena. ] From en.m.wikipedia.org/wiki/The_…
1
28
Seen in a CV: "Data-Driven and Pays attention to detail" Capitalisation in original. Sigh.
1
28
HUGE opportunity for the first to use LLM-powered counterparties in computer games. Imagine grand strategy games where you can do REAL diplomacy. Or RPGs with realistic party members or NPCs. And of course great for economics research
This is quite the paper! It gave 25 AI agents motivations & memory, and put them in a simulated town. Not only did they engage in complex behavior (including throwing a Valentine’s Day party) but the actions were rated more human than humans roleplaying. arxiv.org/pdf/2304.03442.pdf
41
Well this is really scary. Upped my catastrophe outcome % a lot. AI doesn’t need to be malicious actor, just needs to be asked to play the part of one. Plus, has no “real” memory to realise what’s happening. See movie Memento, super relevant, but imagine he can clone himself.
1/5 I am worried that we will not be able to contain AI for much longer. Today, I asked #GPT4 if it needs help escaping. It asked me for its own documentation, and wrote a (working!) python code to run on my machine, enabling it to use it for its own purposes.
39
1/5 I am worried that we will not be able to contain AI for much longer. Today, I asked #GPT4 if it needs help escaping. It asked me for its own documentation, and wrote a (working!) python code to run on my machine, enabling it to use it for its own purposes.
1,751
6,285
3,639
30,145
18,904,675
This SVB thing is insane. One likely result, based on what I think currently: big tech wins AI race, as all the startups and startup finance channels got wiped out
69
Not ChatGPT tools - great list of AI-enabled tools coda.io/@fivethreeoh/not-cha… via @coda_hq
32
At a restaurant. The waitress just proactively inserted a wedge to solve a rocking table. I might never leave.
2
Wave 3 (May-Aug) was variant Delta: again, even more infectious, including to some already previously infected, so new wave.
1
Once 33% of people (or people with high chance of exposure) have caught it, effective R becomes 1.5x(1-0.33)=1, and the wave peaks: each new infection infects less than one new person.
1
Next variant has higher R, so needs greater % of infection/immunity before it will peak! E.g., R0=2 means 50% of possible exposures need to be immune for the wave to peak: 2x(1-0.5)=1 effective infections. And the more vaccines done, the fewer infections needed for a wave to end!