Miguel Martinez
@miguelmalvarez.bsky.social
Director of Innovation and AI, Law Business Research
Co-founder & NED of Signal AI
Co-founder & NED of Signal AI
Reposted by Miguel Martinez
One way AI might help decision-making is by presenting strong positive and negative cases for a policy (“steelmen”) so we can see both sides of an issue.
For example, these are small sections of two OpenAI Deep Research reports on “the case for/against regulating AI development”
For example, these are small sections of two OpenAI Deep Research reports on “the case for/against regulating AI development”
February 23, 2025 at 12:30 AM
One way AI might help decision-making is by presenting strong positive and negative cases for a policy (“steelmen”) so we can see both sides of an issue.
For example, these are small sections of two OpenAI Deep Research reports on “the case for/against regulating AI development”
For example, these are small sections of two OpenAI Deep Research reports on “the case for/against regulating AI development”
Reposted by Miguel Martinez
These are all 100% AI generated Eagles victory celebrations with veo 2, literally all from the first two tries with a text prompt, no editing.
These systems are getting pretty impressive (though you can tell they are AI because no one is trying to climb a greased pole)
These systems are getting pretty impressive (though you can tell they are AI because no one is trying to climb a greased pole)
February 12, 2025 at 6:11 AM
These are all 100% AI generated Eagles victory celebrations with veo 2, literally all from the first two tries with a text prompt, no editing.
These systems are getting pretty impressive (though you can tell they are AI because no one is trying to climb a greased pole)
These systems are getting pretty impressive (though you can tell they are AI because no one is trying to climb a greased pole)
Reposted by Miguel Martinez
I find it pretty amazing that almost every home in Madrid has a metro station within 300 meters on average, and only 25% are more than 500 meters away. Across the entire city! Data from Idealista on @graphext.bsky.social
February 12, 2025 at 5:18 PM
I find it pretty amazing that almost every home in Madrid has a metro station within 300 meters on average, and only 25% are more than 500 meters away. Across the entire city! Data from Idealista on @graphext.bsky.social
I think the genie methaphore as someone who will grant your wish, but potentially twist it completely in ways you didn't expected, is one of the best ones I have heard. It resonates better than the RA one for me
Increasingly phrasing my requests to Reasoner AI models like wishes to a genie: ask for exactly what you want very clearly, with examples if possible, specify everything, no loopholes that can be misinterpreted. (Boring version: request like you would from a RA)
Otherwise you might waste 10 minutes
Otherwise you might waste 10 minutes
February 7, 2025 at 2:25 PM
I think the genie methaphore as someone who will grant your wish, but potentially twist it completely in ways you didn't expected, is one of the best ones I have heard. It resonates better than the RA one for me
Reposted by Miguel Martinez
Less than 3 weeks to get your talks and tutorials in for PyData London 2025!
Submit your proposal by 24 Feb pydata.org/london2025
Submit your proposal by 24 Feb pydata.org/london2025
February 4, 2025 at 2:45 PM
Less than 3 weeks to get your talks and tutorials in for PyData London 2025!
Submit your proposal by 24 Feb pydata.org/london2025
Submit your proposal by 24 Feb pydata.org/london2025
Reposted by Miguel Martinez
wrote about Deep Research, which is very, very good at doing nuanced and complex research.
It is also the first narrow agent that can do sophisticated and likely quite economically valuable work, which tells us something important about the future. open.substack.com/pub/oneusefu...
It is also the first narrow agent that can do sophisticated and likely quite economically valuable work, which tells us something important about the future. open.substack.com/pub/oneusefu...
The End of Search, The Beginning of Research
The first narrow agents are here
open.substack.com
February 3, 2025 at 3:54 PM
wrote about Deep Research, which is very, very good at doing nuanced and complex research.
It is also the first narrow agent that can do sophisticated and likely quite economically valuable work, which tells us something important about the future. open.substack.com/pub/oneusefu...
It is also the first narrow agent that can do sophisticated and likely quite economically valuable work, which tells us something important about the future. open.substack.com/pub/oneusefu...
Reposted by Miguel Martinez
Worthy Read: Strong Opinions, Weakly Held by Abraham Thomas
"Strong opinions weakly held is the optimal strategy if what you care about is rate of learning" 🎯
Want to make better decisions when you don't have all the information?
"Strong opinions weakly held is the optimal strategy if what you care about is rate of learning" 🎯
Want to make better decisions when you don't have all the information?
February 1, 2025 at 2:13 PM
Worthy Read: Strong Opinions, Weakly Held by Abraham Thomas
"Strong opinions weakly held is the optimal strategy if what you care about is rate of learning" 🎯
Want to make better decisions when you don't have all the information?
"Strong opinions weakly held is the optimal strategy if what you care about is rate of learning" 🎯
Want to make better decisions when you don't have all the information?
Reposted by Miguel Martinez
The US Copyright office has ruled that AI/human combined work can be copyrighted as long as a human is adding, changing or selecting elements. Prompts alone do not usually produce copyrighted work. Everything is case-by-case, but the report is clear and thoughtful. copyright.gov/ai/Copyright...
January 30, 2025 at 1:28 PM
The US Copyright office has ruled that AI/human combined work can be copyrighted as long as a human is adding, changing or selecting elements. Prompts alone do not usually produce copyrighted work. Everything is case-by-case, but the report is clear and thoughtful. copyright.gov/ai/Copyright...
Reposted by Miguel Martinez
Every few months, I write an opinionated guide for general purpose users about which AI to pick, especially for newcomers.
Here is my brand new one, which I actually had to update multiple times in the few days I was writing it. Things are changing fast. open.substack.com/pub/oneusefu...
Here is my brand new one, which I actually had to update multiple times in the few days I was writing it. Things are changing fast. open.substack.com/pub/oneusefu...
Which AI to Use Now: An Updated Opinionated Guide
Picking your general-purpose AI
open.substack.com
January 26, 2025 at 2:08 PM
Every few months, I write an opinionated guide for general purpose users about which AI to pick, especially for newcomers.
Here is my brand new one, which I actually had to update multiple times in the few days I was writing it. Things are changing fast. open.substack.com/pub/oneusefu...
Here is my brand new one, which I actually had to update multiple times in the few days I was writing it. Things are changing fast. open.substack.com/pub/oneusefu...
Reposted by Miguel Martinez
Trying something new this year: I'm going to stream spaCy and other open-source development and maintenance twice a week!
I hope this will give the community a better inside look while things are in progress. I'll also answer your questions and chat about NLP, OSS and more.
I hope this will give the community a better inside look while things are in progress. I'll also answer your questions and chat about NLP, OSS and more.
January 6, 2025 at 11:17 AM
Trying something new this year: I'm going to stream spaCy and other open-source development and maintenance twice a week!
I hope this will give the community a better inside look while things are in progress. I'll also answer your questions and chat about NLP, OSS and more.
I hope this will give the community a better inside look while things are in progress. I'll also answer your questions and chat about NLP, OSS and more.
Reposted by Miguel Martinez
🗣️ Our full statement in response to Meta ending its third-party fact checking programme in the US: fullfact.org/blog/2025/ja...
January 7, 2025 at 4:50 PM
🗣️ Our full statement in response to Meta ending its third-party fact checking programme in the US: fullfact.org/blog/2025/ja...
Reposted by Miguel Martinez
My 8000-word note on agents: huyenchip.com//2025/01/07/...
1. An AI-powered agent's capability is determined by its tools and its planning ability
2. How to select the best tools for your agent
3. How to augment a model’s planning capability
4. Agent’s failure modes
Feedback is much appreciated!
1. An AI-powered agent's capability is determined by its tools and its planning ability
2. How to select the best tools for your agent
3. How to augment a model’s planning capability
4. Agent’s failure modes
Feedback is much appreciated!
Agents
Intelligent agents are considered by many to be the ultimate goal of AI. The classic book by Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern Approach (Prentice Hall, 1995), defines ...
huyenchip.com
January 7, 2025 at 6:28 PM
My 8000-word note on agents: huyenchip.com//2025/01/07/...
1. An AI-powered agent's capability is determined by its tools and its planning ability
2. How to select the best tools for your agent
3. How to augment a model’s planning capability
4. Agent’s failure modes
Feedback is much appreciated!
1. An AI-powered agent's capability is determined by its tools and its planning ability
2. How to select the best tools for your agent
3. How to augment a model’s planning capability
4. Agent’s failure modes
Feedback is much appreciated!
Reposted by Miguel Martinez
Whoa! That's a nice compliment just before the holiday season.
Turns out that my recording was the most listened to episode of the Talk Python podcast this year!
Turns out that my recording was the most listened to episode of the Talk Python podcast this year!
Announcing the top @talkpython.fm episodes of 2024!
Featuring @koaning.bsky.social, @cferdinandi.bsky.social, Stanislav Zmiev, Wes McKinney, and @davidism.bsky.social.
Massive write-up not just a list, give it read! cc @mkennedy.codes #python
talkpython.fm/blog/posts/t...
Featuring @koaning.bsky.social, @cferdinandi.bsky.social, Stanislav Zmiev, Wes McKinney, and @davidism.bsky.social.
Massive write-up not just a list, give it read! cc @mkennedy.codes #python
talkpython.fm/blog/posts/t...
Top Talk Python Episodes of 2024
We're rapidly closing in on the end of 2024, and it's been fun to dive into the podcast data and see if we can highlight some of our most well-received episodes of the year. In this post, I break down...
talkpython.fm
December 22, 2024 at 8:12 PM
Whoa! That's a nice compliment just before the holiday season.
Turns out that my recording was the most listened to episode of the Talk Python podcast this year!
Turns out that my recording was the most listened to episode of the Talk Python podcast this year!
Reposted by Miguel Martinez
ModernBERT is available as a slot-in replacement for any BERT-like model, with both 139M param and 395M param sizes.
It has a 8192 sequence length, is extremely efficient, is uniquely great at analyzing code, and much more. Read this for details:
huggingface.co/blog/modern...
It has a 8192 sequence length, is extremely efficient, is uniquely great at analyzing code, and much more. Read this for details:
huggingface.co/blog/modern...
December 19, 2024 at 4:45 PM
ModernBERT is available as a slot-in replacement for any BERT-like model, with both 139M param and 395M param sizes.
It has a 8192 sequence length, is extremely efficient, is uniquely great at analyzing code, and much more. Read this for details:
huggingface.co/blog/modern...
It has a 8192 sequence length, is extremely efficient, is uniquely great at analyzing code, and much more. Read this for details:
huggingface.co/blog/modern...
Reposted by Miguel Martinez
We have just updated our long-term data on life expectancy around the world.
👇 The data by continent shows that life expectancy has more than doubled in every world region.
(Here is all our work on life expectancy: ourworldindata.org/life-expectancy)
👇 The data by continent shows that life expectancy has more than doubled in every world region.
(Here is all our work on life expectancy: ourworldindata.org/life-expectancy)
December 8, 2024 at 8:07 PM
We have just updated our long-term data on life expectancy around the world.
👇 The data by continent shows that life expectancy has more than doubled in every world region.
(Here is all our work on life expectancy: ourworldindata.org/life-expectancy)
👇 The data by continent shows that life expectancy has more than doubled in every world region.
(Here is all our work on life expectancy: ourworldindata.org/life-expectancy)
Reposted by Miguel Martinez
For those looking for a real-world dataset to test your data cleaning, analysis, and ML skills, this one is perfect. It’s as messy as real-world data often is! Thanks to Javier for the analysis and shoutout. paezha.github.io/idealista18/
December 5, 2024 at 5:44 PM
For those looking for a real-world dataset to test your data cleaning, analysis, and ML skills, this one is perfect. It’s as messy as real-world data often is! Thanks to Javier for the analysis and shoutout. paezha.github.io/idealista18/
Great resource! It reminds me of a similar one from evidently www.evidentlyai.com/ml-system-de...
December 2, 2024 at 8:22 PM
Great resource! It reminds me of a similar one from evidently www.evidentlyai.com/ml-system-de...