Jason Bell
@j-jason-bell.bsky.social
I work at the University of Oxford. I post about marketing, econ, AI. Trying to be optimistic and nice. I love capitalism.
“Man,” said the Ghost, “[…] It may be, that in the sight of Heaven, you are more worthless and less fit to live than millions like this poor man’s child. Oh God! To hear the Insect on the leaf pronouncing on the too much life among his hungry brothers in the dust.”
December 25, 2024 at 7:17 AM
“Man,” said the Ghost, “[…] It may be, that in the sight of Heaven, you are more worthless and less fit to live than millions like this poor man’s child. Oh God! To hear the Insect on the leaf pronouncing on the too much life among his hungry brothers in the dust.”
Reposted by Jason Bell
Is psychology going to Cincinnati?
OR: The mystery of the televised salad
www.experimental-history.com
December 7, 2024 at 8:20 PM
Bluesky needs a bookmark feature
December 6, 2024 at 11:58 AM
Bluesky needs a bookmark feature
Interesting but the claim at the end is definitely not justified. This is noisy, no controls, observational, the blue slope might still be negative, and probably non-trivial measurement error.
There's a strong inverse relationship between median household income and zero-car households in Chicago at the tract level, unless you live within a half-mile of an 'L' station.
In other words, rich people will ditch their cars if they can walk to the train.
In other words, rich people will ditch their cars if they can walk to the train.
December 5, 2024 at 4:15 PM
Interesting but the claim at the end is definitely not justified. This is noisy, no controls, observational, the blue slope might still be negative, and probably non-trivial measurement error.
You know uncanny valley. Bryan Johnson has introduced us to uncanny youth.
www.standard.co.uk/lifestyle/we...
www.standard.co.uk/lifestyle/we...
Bryan Johnson: meet the man who is ageing backwards
The multi-millionaire tech entrepreneur Bryan Johnson has made himself a guinea pig for pioneering longevity research. He talks to Hamish MacBain about how he escaped a life of depression to lead the ...
www.standard.co.uk
December 5, 2024 at 9:27 AM
You know uncanny valley. Bryan Johnson has introduced us to uncanny youth.
www.standard.co.uk/lifestyle/we...
www.standard.co.uk/lifestyle/we...
This is the second time Seth says something misleading in the same thread. Isaacman has a degree and it is arguably relevant. There is a third case in the thread so Seth is blocked for me. I hate cronyism but crying wolf is counterproductive.
December 5, 2024 at 7:50 AM
This is the second time Seth says something misleading in the same thread. Isaacman has a degree and it is arguably relevant. There is a third case in the thread so Seth is blocked for me. I hate cronyism but crying wolf is counterproductive.
Nature shouldn’t have this category of article at all
📣 New: AI depends on immense resources, mostly controlled by large companies.
This means "open" AI… isn't very open.
📄 From me, @meredithmeredith.bsky.social and @smw.bsky.social in Nature:
www.nature.com/articles/s41...
This means "open" AI… isn't very open.
📄 From me, @meredithmeredith.bsky.social and @smw.bsky.social in Nature:
www.nature.com/articles/s41...
December 5, 2024 at 7:32 AM
Nature shouldn’t have this category of article at all
I think it’s misleading to say Elon picked the head of NASA. Unless there is extra information that Seth has chosen not to share, it appears Trump chose Isaacman.
December 5, 2024 at 7:29 AM
I think it’s misleading to say Elon picked the head of NASA. Unless there is extra information that Seth has chosen not to share, it appears Trump chose Isaacman.
An SMJ paper I'm on is now out
(1) LLM evals of biz models are closer to expert evals than human non-experts (2) LLM gets closer to experts with diversity (models, personas, prompts) and scaling (more repetitions). Scaling effect > diversity effect
onlinelibrary.wiley.com/doi/full/10....
(1) LLM evals of biz models are closer to expert evals than human non-experts (2) LLM gets closer to experts with diversity (models, personas, prompts) and scaling (more repetitions). Scaling effect > diversity effect
onlinelibrary.wiley.com/doi/full/10....
Generative artificial intelligence and evaluating strategic decisions
Research Summary
Strategic decisions are uncertain and often irreversible. Hence, predicting the value of alternatives is important for strategic decision making. We investigate the use of generativ.....
onlinelibrary.wiley.com
November 22, 2024 at 4:19 PM
An SMJ paper I'm on is now out
(1) LLM evals of biz models are closer to expert evals than human non-experts (2) LLM gets closer to experts with diversity (models, personas, prompts) and scaling (more repetitions). Scaling effect > diversity effect
onlinelibrary.wiley.com/doi/full/10....
(1) LLM evals of biz models are closer to expert evals than human non-experts (2) LLM gets closer to experts with diversity (models, personas, prompts) and scaling (more repetitions). Scaling effect > diversity effect
onlinelibrary.wiley.com/doi/full/10....
I agree with Oleg on all points here. Still think the Atlantic article is worth a read.
Academic intergrity is an important and difficult topic to cover well. But I think the gossipy speculative one-dimensional tone of this article is not constructive, does a poor job of conveying what is actually going on & undermines the important topic being covered. A long thread w/ some examples:
In 2023, Harvard Business School suspended a star professor over charges of research misconduct.
As it happens, one of her papers also had funny data from a *different* B-school superstar -- a "mad, fraudulent unicorn,” per @jamesheathers.bsky.social... (1/3)
www.theatlantic.com/magazine/arc...
As it happens, one of her papers also had funny data from a *different* B-school superstar -- a "mad, fraudulent unicorn,” per @jamesheathers.bsky.social... (1/3)
www.theatlantic.com/magazine/arc...
November 20, 2024 at 11:02 AM
I agree with Oleg on all points here. Still think the Atlantic article is worth a read.
Worth a read
In 2023, Harvard Business School suspended a star professor over charges of research misconduct.
As it happens, one of her papers also had funny data from a *different* B-school superstar -- a "mad, fraudulent unicorn,” per @jamesheathers.bsky.social... (1/3)
www.theatlantic.com/magazine/arc...
As it happens, one of her papers also had funny data from a *different* B-school superstar -- a "mad, fraudulent unicorn,” per @jamesheathers.bsky.social... (1/3)
www.theatlantic.com/magazine/arc...
The Fraudulent Science of Success
Business schools are in the grips of a scandal that threatens to undermine their most influential research—and the credibility of an entire field.
www.theatlantic.com
November 20, 2024 at 8:35 AM
Worth a read
Did anybody do an event study on peaches from Georgia’s bad crop yield I wonder
November 19, 2024 at 6:51 PM
Did anybody do an event study on peaches from Georgia’s bad crop yield I wonder
Not saying I don’t believe this but why not do it blinded?
Didja read about the study that claimed ChatGPT produced poetry “indistinguishable” from Shakespeare?
Critique by NYU’s Ernest Davis, on the unbearable banality of ChatGPT’s poetry: https://cs.nyu.edu/~davise/papers/GPT-Poetry.pdf
Critique by NYU’s Ernest Davis, on the unbearable banality of ChatGPT’s poetry: https://cs.nyu.edu/~davise/papers/GPT-Poetry.pdf
November 19, 2024 at 6:50 PM
Not saying I don’t believe this but why not do it blinded?
A chatbot that assesses aptitude for a variety of tasks based on interviews and produces a report.
Someone is probably building this already.
Someone is probably building this already.
Crazy interesting paper in many ways:
1) Voice-enabled GPT-4o conducted 2 hour
interviews of 1,052 people
2) GPT-4o agents were given the transcripts & prompted to simulate the people
3) The agents were given surveys & tasks. They achieved 85% accuracy in simulating interviewees real answers!
1) Voice-enabled GPT-4o conducted 2 hour
interviews of 1,052 people
2) GPT-4o agents were given the transcripts & prompted to simulate the people
3) The agents were given surveys & tasks. They achieved 85% accuracy in simulating interviewees real answers!
November 19, 2024 at 1:31 PM
A chatbot that assesses aptitude for a variety of tasks based on interviews and produces a report.
Someone is probably building this already.
Someone is probably building this already.
People are saying AI has hit a wall. Specifically scaling transformers. I think it this has been obvious for a while. The big question for me is when we enter the steep part of the next s-curve.
Meanwhile some insane tech is hitting "plateau of enlightenment" stage on the hype cycle. AVs.
Meanwhile some insane tech is hitting "plateau of enlightenment" stage on the hype cycle. AVs.
November 19, 2024 at 1:21 PM
People are saying AI has hit a wall. Specifically scaling transformers. I think it this has been obvious for a while. The big question for me is when we enter the steep part of the next s-curve.
Meanwhile some insane tech is hitting "plateau of enlightenment" stage on the hype cycle. AVs.
Meanwhile some insane tech is hitting "plateau of enlightenment" stage on the hype cycle. AVs.