Links for 2023-03-18
Want to give your agent access to 20k+ tools? “With this integration, you have access to the 5k+ apps and 20k+ actions on Zapier's platform through a natural language API interface. This is extremely powerful and gives your LangChain agents seemingly limitless possibilities.” https://blog.langchain.dev/langchain-zapier-nla/
GPT-4 Khan Academy In Depth Demo https://youtu.be/rnIgnS8Susg
Python-based compiler achieves orders-of-magnitude speedups https://news.mit.edu/2023/codon-python-based-compiler-achieve-orders-magnitude-speedups-0314
US Air Force eyes fleet of 1,000 drone wingmen as planning accelerates https://www.defensenews.com/air/2023/03/08/us-air-force-eyes-fleet-of-1000-drone-wingmen-as-planning-accelerates/
“Here's a 1 paragraph ChatGPT prompt you can use to generate infinite interior design/architecture photographs w/ 90%+ coherence to the prompt in Midjourney” https://twitter.com/nickfloats/status/1635116672054079488
The first jailbreak for ChatGPT-4 that gets around the content filters every time https://twitter.com/alexalbert__/status/1636488553457913856
Speak Foreign Languages with Your Own Voice: Cross-Lingual Neural Codec Language Modeling https://vallex-demo.github.io/
“PwC announces strategic alliance with Harvey, positioning PwC’s Legal Business Solutions at the forefront of legal generative AI” https://www.pwc.com/gx/en/news-room/press-releases/2023/pwc-announces-strategic-alliance-with-harvey-positioning-pwcs-legal-business-solutions-at-the-forefront-of-legal-generative-ai.html
The big idea: should governments run more experiments? https://www.theguardian.com/books/2023/mar/06/the-big-idea-should-governments-run-more-experiments
"...there are 4 basic types (of which our "ordered" solar system is the rarest), dictated by the initial conditions for planetary formation." https://scitechdaily.com/astronomers-show-there-are-four-classes-of-planetary-systems/
The Invisible Graveyard of Crime: A cost-benefit approach to the crackdown in El Salvador https://richardhanania.substack.com/p/the-invisible-graveyard-of-crime
What do British Millennials and Zoomers mean by “socialism”? — “In Britain, the opposite is true today. We get the reverse misunderstanding, because the term “socialism” means something more radical to its supporters than it does to its opponents.” https://iea.org.uk/what-do-british-millennials-and-zoomers-mean-by-socialism/
“…the decline in fertility took hold in France in the 1760s, more than a century earlier than in any other country…regions that secularized experienced a much earlier decline in fertility than those that did not.” https://worksinprogress.co/issue/frances-baby-bust
“Seoul's (0.63) and Shanghai's (0.69) total fertility rate might even be the lowest of any region in the world.” https://twitter.com/landgeist/status/1615342047044435968
Apple wants to move its manufacturing out of China https://qz.com/apple-vietnam-china-foxconn-1850119649
Since there is some confusion about the Monty Fall Problem:
It's easy to see that the answer must be different from the original Monty Hall Problem simply because the game's mechanics are different. In the original, the host always chooses a door with a goat, while in the variant the host chooses randomly.
Intuition:
The Monty Fall Problem is isomorphic to a game with no host and two players that are assigned two different doors at random. Switching cannot possibly be the winning move for both of them. By symmetry, this implies that switching does not increase the probability of winning a car.
It may help to walk through the difference between the Monty *H*all and Monty *F*all problem step by step.
Setup (both variants):
- There are 100 doors.
- The car is behind a random door.
- The player always picks door number 1.
- The host opens 98 of the remaining 99 doors.
- The player is offered to switch to the remaining door.
- The game is repeated N times for large N.
Setup (Monty *H*all):
- The host is told where the car is.
Setup (Monty *F*all):
- The host doesn't know where the car is.
Here is what happens in each variant.
Case 1: The car is behind one of the doors numbered 2 to 100. This happens in 99 out of 100 games.
Monty *H*all:
The host opens 98 of the remaining 99 doors that contain a goat and leaves the door with the car closed.
Monty *F*all:
The host randomly opens 98 of the remaining 99 doors. Because the host chooses randomly, the door with the car is revealed in 98 of 99 games. In only 1 of 99 games, the car remains behind the closed door.
Case 2: The car is behind the door which the player picked, door number 1.
In this case, the hosts in both variants open 98 doors with goats and leave another door, also containing a goat, closed.
We can now simply look at the cases in which switching is an option.
Monty *H*all:
The car is never revealed. The player can always switch. In 99 of 100 games, the car is behind the remaining door that the host has not opened. Therefore, the probability of winning the car by switching is 99/100.
Monty *F*all:
If the car has been revealed, the probability of winning the game by switching is 0 because the player knows for certain that both door 1 and the remaining closed door have goats behind them.
There are just 2 out of 100 games in which the car has not been revealed. In one of those games, the car is behind the remaining door that the host has not opened. In the other game, the car is behind door 1 that the player has chosen. But there is no way for the player to discern these possibilities. Therefore, the probability of winning the car by switching is the same as the probability of winning the car by not switching.