pesenti

Interactive visualization with free access: https://citiesmoving.com/visualizations/

What will the cost be? When sending back function calls results, what will be the number of tokens? Just the ones corresponding to the results or that plus the full context?

Code: https://github.com/facebookresearch/diplomacy_cicero

Blog: https://ai.facebook.com/blog/cicero-ai-negotiates-persuades-...

Site: https://ai.facebook.com/research/cicero/

Expert player vs. Cicero AI: https://www.youtube.com/watch?v=u5192bvUS7k

RFP: https://ai.facebook.com/research/request-for-proposal/toward...

Paper: https://www.science.org/doi/10.1126/science.ade9097

Code: https://github.com/facebookresearch/diplomacy_cicero

Site: https://ai.facebook.com/research/cicero/

Expert player vs. Cicero AI: https://www.youtube.com/watch?v=u5192bvUS7k

RFP: https://ai.facebook.com/research/request-for-proposal/toward...

The most interesting anecdote I heard from the team: "during the tournament dozens of human players never even suspected they were playing against a bot even though we played dozens of games online."

Research paper: https://makeavideo.studio/Make-A-Video.pdf

Examples: https://make-a-video.github.io/

Demo site: https://makeavideo.studio/

I am told live demo and open model are on the way.

Research paper: https://makeavideo.studio/Make-A-Video.pdf

Examples: https://make-a-video.github.io/

Demo site: https://makeavideo.studio/

I am told live demo and open model are on the way.

Sure, agreed, I try to always submit the exact title for that reason but I got lazy as I was on my phone.

I do.

I am actually no longer working for Meta. I do think the title reflects the news and is factually correct. But happy for @dang to change it back as it’s not exactly the title of the article

Are all the 200x200 translations going directly or is English (or another language) used as an intermediate for some of them?

https://datascience.stackexchange.com/questions/62868/high-l...

Blog post: https://ai.facebook.com/blog/nllb-200-high-quality-machine-t...

Paper: https://research.facebook.com/publications/no-language-left-...

Github: https://github.com/facebookresearch/fairseq/tree/nllb/

Our intelligence isn't task specific, but that doesn't mean it can solve any problem. It's actually full of biases and very optimized for our survival (vs being a general problem solver). It's ok to talk about more or less narrow/general tasks/intelligence. But what threshold of generality is "general"?

And the problem is that once people assert this "absolute" level of generality, they assume it can do anything, including make itself more intelligent.

AGI is a concept that people throw around without defining. Human intelligence is not general, it's highly specialized. So if AGI is not human-level intelligence, what is it? https://twitter.com/ylecun/status/1204013978210320384?s=20&t...

Legality aside, this is a terrible and naive way to think of DEI. Candidates recruited this way will have a stigma attached to them (“they were recruited because they were X”).

A better way is to enforce DSA - Diverse Slate Approach - which ensure that you don’t recruit anybody without having considered multiple genders and/or ethnicity before making the offer.

The goal of DSA is to recruit the best candidate for the role, but ensure a broad set of candidates has been considered. It’s a win win for all and it works.

Which feature are you talking about? This is a research project, not a product. The point of research projects is to investigate and figure out answers we don't have today.

It's hard to elaborate on something that's not developed yet.

Here are the principles we are using when developing these products:

https://about.facebook.com/realitylabs/responsible-innovatio...

In practice, it will mean that some things you will know about, but don't necessarily give consent (i.e., someone taking a picture or a video of you), while other will likely require consent (i.e., recognizing you in these pictures/videos).

No, that’s not a reasonable assumption. Photos or videos of you are likely on Facebook’s servers, uploaded by friends or others when you are in public.

What I can assure you though is that if you are not a user, we are not recognizing you in these pictures and videos. We won’t do that without explicit consent from users.

These are great points. And yes, we'll only get there if it becomes part of the culture, which this is trying to communicate:

https://about.facebook.com/realitylabs/responsible-innovatio...

but I don't expect you to take it at face value yet, we have a long way to go.

Here is a link to all the tools we provide: https://research.fb.com/data/

From https://about.facebook.com/realitylabs/projectaria/

"Participants will only record in either Facebook offices (once they reopen), wearers’ private homes (with consent from all members of the household), or public spaces, and won’t record in private venues without written consent from such places. Before any information gathered in a public place is made available to our researchers, it will be automatically scrubbed to blur faces and vehicle license plates."

So we anonymize all the content collected in public. I don't have the stats for the face blurring algorithm, and while you can ask for your data to be removed, we don't provide locations/times. These are good suggestions though.

The question was about opting out of our "utopia", I wouldn't call that a "straightforward question". If you do have a straightforward question, especially about this or another specific project, I can try to provide a straightforward answer.

My point in linking to the settlement wasn't about the fine, it was about the guarantees, oversight, and accountability that came with it. Read the FTC news release, and you'll see that these are pretty extensive.

And I agree completely on the public conversation, this is why we released this dataset and why we are doing project Aria.

That’s a good question and I don’t think we have all the answers yet. But I expect that many functionalities will need to request and get consent from 3rd parties.

No but they should:

https://www.ftc.gov/news-events/press-releases/2019/07/ftc-i...

I encourage you to read the details of the news, in particular how the data was collected with full consent from all (something that most research datasets don’t do well). It doesn’t answer your question directly but gives you an idea on how we think about these issues. AI-powered wearables will have stringent privacy guarantees, both for the wearer and the people around.

I support Facebook AI. The actual news is that we are releasing an extensive egocentric dataset with associated tasks for the research community. Collecting such dataset in a private and responsible way is a significant challenge. Happy to answer any questions.

https://ai.facebook.com/blog/teaching-ai-to-perceive-the-wor...

Our main usages of speech rec are:

- Understanding for videos for content moderation, ranking and recommendations

- Captioning

- Assistant tech in Portal, Oculus and upcoming glasses

The communication around this change was very poorly handled. This policy update does not affect the privacy of messages with friends or family in any way: https://faq.whatsapp.com/general/security-and-privacy/answer.... I don't believe that was well understood.

Disclosure: I work at FB but unrelated to WhatsApp

Facebook employee here. You can definitely continue talking about any company business or decisions, even if they have political relevance, as you did before. This change only applies to discussions completely unrelated to work.