Will a multimodal LLM be able to generate instructions for origami crafts by 2030? | Manifold

Will a multimodal LLM be able to generate instructions for origami crafts by 2030?

Basic

6

Ṁ230

2030

83%

chance

1D

1W

1M

ALL

Will resolve to YES if a comment describing a prompt format, left here for a given accessible LLM, manages to generate correct instructions to make a 3D origami figure for any given animal at least 75% of the time. The instructions can be in any form (text, image, speech, video...), as long as they are unambiguous, possible, and achieve the correct shape (admittedly this can be a bit subjective, hence the 75% allowance).

This question is managed and resolved by Manifold.

#️ Technology

#Technical AI Timelines

Get

1,000

and

3.00

Sort by:

The trouble I see here is that I don't expect that AI models that fill the role of ChatGPT and similar in the year 2030 will be referred to as "LLMs". They will model language but only as a subset of their more important task of general world modeling. So the question seems a bit off.

@JohnEg GPT-4 is already multimodal, not modelling only language, but we still widely call it an LLM. I really wish you end up right and research does move more towards "general world modelling" instead of larger and larger models on more data, but I am pessimistic. Ultimately you did interpret the spirit of the question correctly: "AI models that fill the role of ChatGPT and similar in the year 2030" so I don't think an urgent fix is needed on the question.

Related questions

Will one of the major LLMs be capable of continual lifelong learning (learning from inference runs) by EOY 2025?

-16% 1d34% chance

By 2029 end, will it be generally agreed upon that LLM produced text/code > human text/code for training LLMs?

Will "large reasoning models" significantly outperform traditional LLMs in creative writing by 2026?

By 2028 will we be able to identify distinct submodules/algorithms within LLMs?

Will there be any simple text-based task that most humans can solve, but top LLMs can't? By the end of 2026

Will I write an academic paper using an LLM by 2030?

[Situational awareness] Will pre-2026 LLMs achieve token-output control?

In 2030, will LLMs be sending messages to coordinate with one another, whether we can decode them or not?

By 2025 end, will it be generally agreed upon that LLM produced text/code > human text/code for training LLMs?

LLMs to the moon? (By 2030)

Related questions

Will one of the major LLMs be capable of continual lifelong learning (learning from inference runs) by EOY 2025?

Will I write an academic paper using an LLM by 2030?

By 2029 end, will it be generally agreed upon that LLM produced text/code > human text/code for training LLMs?

[Situational awareness] Will pre-2026 LLMs achieve token-output control?

Will "large reasoning models" significantly outperform traditional LLMs in creative writing by 2026?

In 2030, will LLMs be sending messages to coordinate with one another, whether we can decode them or not?

By 2028 will we be able to identify distinct submodules/algorithms within LLMs?

By 2025 end, will it be generally agreed upon that LLM produced text/code > human text/code for training LLMs?

Will there be any simple text-based task that most humans can solve, but top LLMs can't? By the end of 2026

LLMs to the moon? (By 2030)

© Manifold Markets, Inc.•Terms + Mana-only Terms•Privacy•Rules