Once introduced as “defender of the null hypothesis”. Founder. PhD in Evo Psych. Writing: @deepdebates
Toronto
Joined October 2010
- Tweets3.4K
- Following3.4K
- Followers818
- Likes11.9K
Pinned Tweet
Those studying AGI refuse to consider that the mind evolved to be highly specialized with massive ‘domain-specific’ innate circuitry.
They refuse because they think evolved specialization is antithetical to GENERAL intelligence.
In fact, this highly specialized circuitry is precisely what enables general intelligence.
It’s counterintuitive, but computational constraints enable general (broad) problem solving.
It turns out you need to hire politicians that bring in the woke environmental politics that in turn give the sales ppl the moral high ground to replace paper with a worse, but environmentally friendly product.
30
Shea Balish retweeted
if you ever happen to find yourself shitting in your own car
because you are mad at elon musk about politics
consider for a moment you may be mentally ill
Scott Galloway says he owned a Tesla, but sold it because of Elon.
"Not before taking a giant dump in the passenger seat. True story."
He's gone from suggesting Elon be executed to telling stories of taking dumps in a car he owned because it was made by Elon's company.
This is a broken, deranged and utterly toxic individual.
Shea Balish retweeted
it's hilarious how many have taken the bitter lesson to mean just scaling up transformers. it's about scaling, but innovations in algorithmic primitives can lead to much more efficient scaling
Shea Balish retweeted
Data scaling and compute scaling are *multipliers* for the generalization power of an approach. Trillions of dollars in investment can make very inefficient, low-generalization approaches look adequate, but that doesn't make them the ultimate end-all-be-all of AI.
You should work on methods that *can* scale, because if you don't benefit from this multiplier you simply can't compete. But you should keep in mind scaling is only a multiplier: a cleverer approach will scale *better*.
Shea Balish retweeted
Great talk from Ken on how to understand and measure open-ended systems
ARC Prize Research Summit - @kenneth0stanley
On why open-endedness, invention, discovery are essential to AGI, but designing a benchmark for them creates difficult problems.
Shea Balish retweeted
Final form is reasoning in the weights, knowledge outside, in space, via solar, distributed training, continually learns, and aligned
Shea Balish retweeted
"AI is just matrix multiplication" is the new "emotions are just chemicals in the brain"
Shea Balish retweeted
What if there is no hard problem? @Plinz
Wisdom, understood as a deeply contextual and pragmatic awareness of human nature, requires more than brute-force gradient descent of natural language.
I find AI to be extremely knowledgeable, and so exceptionally able to detect references to almost anything, but not very wise, and so it only makes the most surface-level judgements about what point some writing is making.
This is particularly true when it's analyzing writing that it just read for the first time, as opposed to analyzing writing that's already in its training data (along with lots of analysis of that writing by other people)
AI's knowledgeableness is easier to see than its limited wisdom; it's perhaps even the very definition of wisdom that it cannot be easily verified (as opposed to intelligence).
1
87
In 2022, I asked @WillManidis at @hereticon what he thinks of @nachkari: Will said he’s likely to go down as one of the best investors ever. True story.
41
Shea Balish retweeted
I'm far less concerned about a world ruled by today's Chinese than I was about a world ruled by the old USSR. So I'm far less invested in conflict w/ China today than I was in the old cold war.
Shea Balish retweeted
When a phone rings at a funeral, people don't just silence it. They make a show of turning it off. That performance is embarrassment (Feel #53) at work: telling the room the rule was broken by accident. @ChrisWillx and I got into this on Modern Wisdom. Out now: podcasts.apple.com/us/podcas…
