Eh, it appears that the Microsoft and other investors want Sam Altman back, but the employees are the ones that are making it happen. It appears that Sam was fired because of AI safety reasons but everyone who is really really concerned about that kind of thing already left and works for Anthropic, so Ilya’s “But AI Safety” doesn’t come off as very convincing, especially with the way Sam was fired. I suspect that people work at OpenAI because they want to, not because they have to. I mean, what were they expecting?
glagidse
Karma: 11
For me, Vim isn’t about speed, it’s about staying focused. If I have to move my hand between the keyboard and the mouse every time I want to move the cursor, I lose a little bit of focus. Vim solves this issue by not having to move the placement of your hands. Of course a pointing stick also solves most of these issues.
The way I understood the intuition behind the Monty hall problem is so:
1. You’ve got a million doors
2. Only one door has a prize
3. Imagine what the probability is that you pick one at random (one in a million)
4. Pick one at random
5. Randomly open 99998 doors that you did not pick and do not contain prizes.
6. 2 doors are remaining, one of which you picked.
7. Has the probability that you picked the correct door changed?
8. If yes, why yes? and if no why no? And what is the new probability?
Indeed, it seems to me that Robin Hanson isn’t for making blackmail legal, but for making blackmail legal plus a bunch of extra rules without explicitly describing what those extra rules should be.
Blackmail plus a bunch of rules is not blackmail as most people understand it I think.
Sure, but the first time you told that lie you probably used system 2 thinking. Your brain might have optimized this process by creating a heuristic and programmed it into system 1, making lying a simple reactionary response. If you’re using only system 1 it’s not conscious deception. You’re not *deciding* anything, it a simple reaction.
As for your example, I think most people with no experience flirt with their system 2, optimizing their thought processes to use the least amount of system 2 as possible. Though you are right that I might have phrased my statement better.
Statement 2 should have been:
Conscious deception uses system 2 thinking. Non conscious deception uses system 1 thinking. Accidental deception uses no thinking (added for completion).
Would be a better statement I think.
My point is that the statement
*The Law of Least Effort Contributes to the Conjunction Fallacy*
Is false. for the reasons mentioned above.
I’ve skimmed over it. But I guess I have not written down my thoughts on how the Conjecture fallacy relates to social behavior.
If the following two statements are true:
1. The conjunction fallacy mistake is made mainly because people overly rely on system 1 thinking.
2. Complex social behavior, like deception requires system 2 thinking.
Then the following statement is obviously false:
3. People make the conjecture fallacy mistake because of complex social behavioral reasons.
I think statement 1 and 2 are true, therefore I think 3 is false. But because I think 3 is false does not mean I think that making the cojecture fallacy mistake has no social implications. Someone who knows the default heuristics programmed into humans has advantages over those that don’t in social situations.
People make the Conjecture fallacy mistake for the same reason as when they read the following question:
“A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost?”
The first thought that enters your mind is $0.10 cents. That’s your system 1 speaking. If you want to figure out the correct awnser you’ll have to use your system 2. Does the fact that system 1 thinks “$0.10” have social implications? sure. Does system 1 think “$0.10″ because of complex social reasons? I doubt it.
Let’s take wikipedia’s example of the conjunction fallacy:
----
Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.
Which is more probable?
A: Linda is a bank teller.
B: Linda is a bank teller and is active in the feminist movement.
----
The majority of people choose the answer B. So you ask, why do the majority of people choose option B? If you explain the conjunction fallacy to them most if not all of them will change their minds and choose option A. So what is going on? Could it be that they’re choosing option B because they actually think that they’re answering a different question? What question are they answering?
Take these possible answers to possible questions:
Question: unknown
A: Rick is a scientist
B: Rick is a scientist and has gray hair
Question: unknown
A: Bella is a pianist
B: Bella is a pianist with glasses
Question: unknown
A: Shaggy has a soulpatch
B: Shaggy has a soulpatch and a green shirt
Now, if B is the correct awnser for all these examples, and you take your Occam’s razor, what can infer about the questions?
Well, B answers have more associations with the questions than the A answers. So if we go back to the first question, why do most people choose B over A? Why do they answer the question “which of these two answers have more associations with the question” instead of the *actual* question being asked? It’s because answering that question requires less energy. It’s the default heuristic that’s programmed into your brain, it only requires system 1 thinking.
Because that’s the default heuristic programmed in by evolution, it’s going to be used by default. To change it, you need to activate your System 2 thinking to update your system 1 thinking to trigger your system 2 thinking when you come across questions like these, this requires a lot of energy. For whatever reason, our evolution prioritized conservation of energy over answering the conjunction questions correctly.
I suspect you and most of the readers here have a “tricky question alarm” trigger programmed in our system 1. When we see questions like these our tricky question alarm goes off and our system 2 activates so we can reason over them. Not everyone has this “tricky question alarm” trigger programmed in their head, and even if you do have such a trigger, it does not always go off.
One way you can verify this is by checking how fast people answer these kind of questions. I suspect that people who answer B instead of A are much faster in giving their answers because they don’t activate their system 2.
It is primarily an energy conservation thing, not a social thing.
Apparent moral: Apparent corrigibility from training or surface behavior doesn’t fix inner goals; without hard guarantees and control, you lose.
Actual moral: First to superintelligence wins; second place is ash.