Apparently they did CT scans of closed books and read the content. Polevoy, Dmitry V., et al. "From tomographic reconstruction to automatic text recognition: the next frontier task for the artificial intelligence." Fifteenth International Conference on Machine Vision (ICMV 2022). Vol. 12701. SPIE, 2023. https://iris.unive.it/bitstream/10278/3687069/1/Albertin_et-...

So yeah, but lottery companies probably make it harder by engineering against it.

And of course you need to test and debug before you ship to production.

Reading, analyzing and assembling documentation could be probably done by LLMs.

And by including old code and snippets into the training set, the LLM could be fairly proficient in writing this code probably too?

Maybe someone knows more about the use/not-use of LLMs in this context?

The problem is: publication is based on reputation. Reputation takes time and effort from the entire community.

I feel like modern infrastructure (Google Scholar, AI research, LinkedIn, etc) helped to decrease the importance of high-impact journals such as Nature, etc. Researchers don't rely on highly curated printed journals in their physical mailbox to get informed what's happening. You can just use tools to scrape content much faster.

But still: It can be career decisive if a reseachers lands a publication in a for-profit journal such as Nature.

The CS community has a much nicer publishing pipeline where most top journals/proceedings are attached to non-profit conferences and the fee is 0 (beside a conference fee).

I wish more fields would work like this: you publish with a conference proceeding and talk on the conference about your paper.

Researchers are themselves responsible for typesetting, advertising, etc. This and removing for-profit stakeholders can reduce the costs a lot.

For me the solution is in signed e-mails and signed documents. If the person invites me to a online meeting with a signed e-mail, I trust that person that it's really them.

Same for footage of wars, etc. The journalist taking it basically signs the videos and verifies it's authenticity. It is AI generated, then we would loose trust in that person and wouldn't use their material anymore.

The operating margin of Elsevier is around 40% which is huge! At the end mostly paid by tax-payer money.

Personally, I never review or publish with Elsevier.

https://news.ycombinator.com/item?id=46132901

So yes, you use it to write the paper but soon it is public knowledge anyway.

I am not sure if there is much to learn from the draft of the authors.

Later at 6:45 they show more people testing them

The problem is the same, the relative concentration of oxygen in air is less than 0.05% (~450pars per million). In water much less.

Recent article: https://www.theguardian.com/environment/2025/nov/28/africa-f...

A computer uses orders of magnitude less energy than a human.

It's all about the task, humans are specialized too.

EDIT: maybe add a logarithm or other non-linear functions to make the gap even bigger.

https://dkriesel.com/en/blog/2013/0802_xerox-workcentres_are...

https://en.wikipedia.org/wiki/Koralm_Tunnel