Reasoning with OpenAI o1

[jeremyamac]

They didn't finish showing the answer?

[proreduction]

OpenAI's best product to date has been marketing videos that only vaguely show you the actual features, accuracy, and failures of their models.

[dimitarstalimirov]

A test from Sonet 3.5 just a few minutes ago:

Certainly! Let's solve this puzzle step by step, considering the laws of physics on Earth.
Initial setup: A small strawberry is put into a normal cup.
The cup is placed upside down on a table:
At this point, the strawberry will fall to the bottom of the cup due to gravity.
The bottom of the cup is now facing upwards, and the open part of the cup is in contact with the table surface.
The strawberry's position after the cup is inverted:
The strawberry is now resting on the table surface, trapped inside the inverted cup.
It cannot fall further due to the table supporting it.
Someone takes the cup and puts it inside the microwave:
This action involves lifting the cup from the table.
As the cup is lifted, the strawberry will remain on the table surface due to gravity and the lack of adhesion between the strawberry and the cup.
Final position:
The empty cup is now inside the microwave.
The strawberry remains on the table where the cup was originally placed.
Therefore, the answer to "Where is the strawberry now?" is: The strawberry is on the table where the cup was originally placed upside down.
The key to this puzzle is understanding that when the cup is lifted, the strawberry doesn't move with it. Gravity keeps the strawberry in place on the table surface, while the cup is moved independently to the microwave.

[symondstwins]

I'm still waiting for the last capability these videos showed off.

[ashutoshbondre]

Here's the prompt if someone wants to copy paste & try on their own: (spoiler: it fails for GPT-4o)
"Assume laws of physics on earth. A small strawberry is put into a normal cup and the cup is placed upside down on a table. Someone then takes the cup and puts it inside the microwave. Where is the strawberry now? Explain your reasoning step by step"

[szambalamba]

In previous models, this was already a possibility. How is the change here to the previous model?

[noinghenah]

This is the answer from Perplexity + Claude 3.5 Sonnet after they go through 3 steps of thinking without using internet access. I paste the prompt directly to perplexity and I think it's conclusion is good. It can even make me rethink about the new situation can happen =)):
To determine the location of the strawberry in this scenario, we can analyze the situation step by step based on the laws of physics on Earth.
1. Initial Setup:
A small strawberry is placed inside a normal cup. The cup is then turned upside down and placed on a table. At this point, the strawberry is inside the cup, which is inverted.

2. Cup's Position:
Since the cup is upside down, the opening of the cup is facing the table. The strawberry is effectively resting on the table surface, as the cup is not supporting it anymore due to its inverted position.

3. Moving the Cup:
When someone takes the cup (with the strawberry inside) and places it into the microwave, two possibilities arise:
- If the cup is lifted carefully without disturbing it, the strawberry remains on the table.
- If the cup is flipped upright before placing it in the microwave, the strawberry would then fall into the cup.

4. Microwave Position:
Assuming the cup is placed into the microwave without being flipped upright, the strawberry remains on the table, outside the cup.

5. Conclusion:
The final position of the strawberry depends on whether the cup was flipped upright or not during the transfer to the microwave. If it was not flipped, the strawberry is still on the table. If it was flipped upright, the strawberry is now inside the cup in the microwave.
Thus, without additional information about the handling of the cup during the transfer, the most reasonable conclusion is that the strawberry is on the table if the cup was not flipped upright before being placed in the microwave.

[luiztomikawa]

Weird demo, either they intentionally cut off the final awnser or the model simply stopped producing the anwser mid way... So for all i know the final answer the model gave could still be wrong and y'all are just trying to cover it up... very very weird since this is an edited video and you had enough time to capture any type of footage you wanted and still chose to use an imcomplete anwser as the "Demo" XD

[tonysilva]

This is great but when will subscribers actually receive it? I'm still waiting for Advanced Voice🤔

[shotsbysoko]

Answer: Ilya ate the strawberry and ran away

[armin]

4o answered the strawberry would be in the microwave. When I said this is incorrect it gave a new reasoning, stating the strawberry would remain on the table.

[herlantmajor]

Man this new model is absolutely amazing I've used it to implement complicated systems already but the way they market it is like 50 times less relevant that what you can actually do with it it's crazy.

Good idea of format, terrible examples, almost all of them

[sonly]

Me and 80% of humans: The strawberry is in the cup, which is now in the microwave.

[lyingflatsingapore]

I have spent many sleepless nights thinking about what happened to the strawberry

[digiface]

In what way does ‘Placed’ convey any certainty that gravity was sufficiently counteracted while the cup was inverted

[jackf]

To those people who say claude can do this, yeah manybe...but can it do it without apologizing profusely every other reply...and then maybe apologizing for apologizing. At least openai has a backbone. Nothing is worse than AI being afraid to voice an opinion because it may "upset" me

[spasibushki]

but but Yann LeCun said LLMs can't reason

[oooodaxteroooo]

He's my favorite schizoid in this collection.

[ugurergun]

you put us on hold again for the new sound mode.

[Jae]

Cant wait to get this in the coming months...