Sign in

Fernando Martel García

@fmgbsky.bsky.social
59 followers 243 following 59 posts

𝙲𝚊𝚞𝚜𝚊𝚕 𝙸𝚗𝚏𝚎𝚛𝚎𝚗𝚌𝚎 | 𝚁𝚎𝚜𝚎𝚊𝚛𝚌𝚑 𝙿𝚛𝚊𝚌𝚝𝚒𝚌𝚎 | 𝙴𝚗𝚝𝚛𝚎𝚙𝚛𝚎𝚗𝚎𝚞𝚛𝚜𝚑𝚒𝚙 𝕎𝕒𝕤𝕙𝕚𝕟𝕘𝕥𝕠𝕟, 𝔻.ℂ. | www.fernandomartel.com

PostsRepliesMedia
Reposted by Fernando Martel García
Joshua Gans @joshgans.bsky.social · 07/10/2026
My thoughts on OpenAI's Big Maths Day. joshuagans.substack.com/p/openai-dro...
joshuagans.substack.com
OpenAI drops a bomb on maths
October 6th 2026 was a big day
033
Fernando Martel García @fmgbsky.bsky.social · 07/10/2026
In sum, the discussion about disclosures of LLMs and concerns about credit is a little bit like farm labourers discussing piece rates during the arrival of the combine harvester. They are trying to salvage the old vs adapting to the new: Seeking point solutions to systemic problems.
100
Fernando Martel García @fmgbsky.bsky.social · 07/10/2026
Back in ~2005 I shared this slide at a conference. I was dismayed how artisanal research production was. I even quit academia to start a company bringing abstraction, standardization, and automation to the research process. LLMs may succeed where I failed. Credit makes sense in an artisanal world...
100
Fernando Martel García @fmgbsky.bsky.social · 07/10/2026
IMHO "qualitative leap" needs unpacking. Credit is a niche, academic concern. Society cares about impact, not how it came about. Besides, as research tasks get automated away, will there be any human credit left? The world of academia is about to be disrupted in fundamental ways.
100
Fernando Martel García @fmgbsky.bsky.social · 07/10/2026
At the other extreme: Why not disclose what was the researcher's contribution? The assumption being that most of the work was done by LLMs. Even then, why make it explicit vs leaving it implicit? If everyone can access similar LLMs differences will still be driven by authors, no?
100
Fernando Martel García @fmgbsky.bsky.social · 07/10/2026
In the spirit of provocation: Should one disclose the use of a calculator, a computer, unit tests? The piece should state briefly what makes LLMs uniquely different from other tech. Besides, soon everyone will use LLMs. Hence, a disclosure might simply state the obvious, be uninformative.
100
Fernando Martel García @fmgbsky.bsky.social · 27/09/2026
la mediocridad está de oferta
010
Fernando Martel García @fmgbsky.bsky.social · 19/07/2026
As William Greene liked to ask in his econometrics class: "This estimator produces an estimate but what is it an estimate of?"
000
Fernando Martel García @fmgbsky.bsky.social · 19/07/2026
A great reference here is Gary Goertz's book "Social Science Concepts". A treatise on how to build concepts that are theoretically and empirically coherent. Concept definition is key to pinning down the estimand. Without it, how can we evaluate the bias variance tradeoff when defining an index?
110
Fernando Martel García @fmgbsky.bsky.social · 13/07/2026
Metaculus employment forecast vs government forecast. One highlight: "unemployment for 4-year college graduates in the 22-27 age range expected to grow from the current 6% to 12.8% in 2035".
unemployment forecast
010
Fernando Martel García @fmgbsky.bsky.social · 11/06/2026
Great piece highlighting the political roots of macro instability. What is the latest on the dynamics of how economic policy and political equilibria interact? The Chilean Mirror: What Argentina Sees and Cannot Yet Reach
marcusnunes.substack.com
The Chilean Mirror: What Argentina Sees and Cannot Yet Reach
Both countries export commodities, both have experienced boom and bust, and both have attempted radical economic reform. The difference between them is not primarily technical. It is institutional.
000
Fernando Martel García @fmgbsky.bsky.social · 20/05/2026
For lovers of the arcane: Knotting Matters Magazine - IGKT
igkt.net
Knotting Matters Magazine - IGKT
International Guild of Knot Tyers
000
Fernando Martel García @fmgbsky.bsky.social · 24/04/2026
Great article by @timoreilly on scenarios for AI adoption and impact on jobs. www.oreilly.com/radar/scenar...
000
Fernando Martel García @fmgbsky.bsky.social · 16/04/2026
As the cost of the marginal robustness check approaches zero, what will happen to the number of robustness checks per paper, and the length of the annexes? What will be the new binding constraint in scientific communication?
030
Fernando Martel García @fmgbsky.bsky.social · 15/04/2026
One aspect worth considering, and hinted at but not discussed explicitly in the article (as far as I could tell), is the value of information. I discuss this briefly in the section "Move fast, break (little) things" in this blog post: www.fernandomartel.com/posts/politi...
fernandomartel.com
Political science as problem solving | Fernando Martel García
010
Fernando Martel García @fmgbsky.bsky.social · 15/03/2026
What do academic researchers and traffic cops have in common? They all work at intersections. 🤷
001
Fernando Martel García @fmgbsky.bsky.social · 13/03/2026
Gated
000
Fernando Martel García @fmgbsky.bsky.social · 12/03/2026
Will AI agents replace software? My experience building an options-trading record-keeping workflow suggests otherwise. AI is useful for predictive tasks. But when a workflow must be exact, stateful, and auditable, software still needs to stay in charge.
fernandomartel.com
Can you trust stochastic agents in deterministic applications? | Fernando Martel García
Today I want to share my experience automating a record-keeping workflow using AI. What makes this application interesting is that financial record-keeping ought to be exact and stateful, yet the LLM…
000
Fernando Martel García @fmgbsky.bsky.social · 11/03/2026
I recently came across a study that claimed llms ignore llm.text. Nothing reads them apparently.
000
Fernando Martel García @fmgbsky.bsky.social · 05/03/2026
I guess AJPS finally qualifies as a scientific journal. Previously AJPS had a formal editorial policy of not even considering for review unsolicited (!?) rejoinders. Not a serious journal.
010
Fernando Martel García @fmgbsky.bsky.social · 05/03/2026
I personally am not a big fan of notebooks but just using AI and reams of md files is not ideal either. Listen to this podcast creators.spotify.com/pod/profile/... Or just search YouTube for any Bret Victor talk.
creators.spotify.com
"Vibe Coding is a Slot Machine" - Jeremy Howard by Machine Learning Street Talk (MLST)
Dive into the realities of AI-assisted coding, the origins of modern fine-tuning, and the cognitive science behind machine learning with fast.ai founder Jeremy Howard. In this episode, we unpack why A...
000
Fernando Martel García @fmgbsky.bsky.social · 04/03/2026
Great podcast on the potential and limitations of LLMs from an expert in the field: "Vibe Coding is a Slot Machine" - Jeremy Howard
podcasters.spotify.com
000
Fernando Martel García @fmgbsky.bsky.social · 03/03/2026
Homeschooled kids can be done with school in 1-2 hours and spend the rest of the day playing with friends, specially if they are similarly home schooled.
100
Fernando Martel García @fmgbsky.bsky.social · 03/03/2026
I home schooled my son during COVID on USA. The most rewarding experience of the time. Sailors often home school their kids. Learnings: * In a small group you can cover most of the school material in 1-2 hours per day, move beyond grade, and get A grades. * school is 80% daycare, 20% learning
000
Fernando Martel García @fmgbsky.bsky.social · 03/03/2026
"Last week, we estimated a 66% chance that the US would strike Iran before April, well above the Polymarket aggregate of 20.5%" - open.substack.com/pub/sentinel...
open.substack.com
🟡 US and Israel strike Iran, Iran retaliates against US allies, Anthropic designated a supply chain risk || Global Risks Weekly Roundup #9/2026
Executive summary
010
Fernando Martel García @fmgbsky.bsky.social · 28/02/2026
One concern with AI is that AI will do the work, cheating, etc... Another concern, perhaps more intriguing, is what students should learn nowadays. My sense is more abstractions (the why and the what) and less implementation details (the how). More math, dags, proofs, less code.
110
Fernando Martel García @fmgbsky.bsky.social · 28/02/2026
Simple beats complex. Maybe not a universal principle but a good point of departure....
020
Fernando Martel García @fmgbsky.bsky.social · 27/02/2026
Love the syllabus. Two things caught my attention. 1. Would love to see a blog post expanding the section on requirements and policies in age of AI. You are in a unique position to witness and shape the impact of AI on education. 2. No Paul Rosenbaum (e.g design of observational studies)?
110
Fernando Martel García @fmgbsky.bsky.social · 26/02/2026
I wrote a new post on automated policy evaluation: using AI to scale policy impact analysis with observational data. This could speed up knowledge production, but whether it improves welfare depends on the institutions that adopt policy.
fernandomartel.com
Automated Policy Evaluation | Fernando Martel García
Today I came across the Project APE, for automated policy evaluation (APE). The project uses Claude Code plus public observational datasets to generate end-to-end economics-style policy impact…
000
Fernando Martel García @fmgbsky.bsky.social · 05/02/2026
Market commentary isn't usually my thing but Microsoft's 15% stock price decline after beating earnings last month raises deeper questions about capital allocation within firms.
fernandomartel.com
Microsoft's Ai Capital Allocation | Fernando Martel García
Microsoft’s earnings report on January 28th beat market expectations yet the stock fell more than 15% soon after. Wall Street was spooked by two revelations in the earnings. First, 45% of Azure’s…
010
Fernando Martel García @fmgbsky.bsky.social · 19/01/2026
LLMs created a weird shift for me. I’m more productive than ever, but that makes free time feel less relaxing, not more. The more I can do, the harder it is to step away.
000
Fernando Martel García @fmgbsky.bsky.social · 09/12/2025
How much of that work do you think can be abstracted, standardized (e.g. checklists), and automated? Peer reviews seem to combine haphazardness with labor intensity. Surely there is room for an orders of magnitude better process.
000
Fernando Martel García @fmgbsky.bsky.social · 09/12/2025
Built FermiApp and open source tool for back-of-the-envelope strategic analysis using Fermi estimation. In my blog post I explain how working with AI as engineering partner (not autonomous agent) helped me move 10x faster while staying in control: www.futureofscience.ai/post/2025/11...
futureofscience.ai
Building FermiApp: AI as Engineering Partner | The Future of Science
Better business outcomes through causal AI insights
010
Fernando Martel García @fmgbsky.bsky.social · 05/11/2025
I find this very counterintuitive. IMHO longer letters send a worse signal than shorter letters. Shorter is a lot harder than longer, while still a packing a punch. Why are folks using AI to write longer letters?
000
Fernando Martel García @fmgbsky.bsky.social · 25/09/2025
Such a metric of success may introduce a conservative incentive, no? E.g. where are you more likely to deviate all else equal: Doing a field experiment in a war torn country, or with grad student at Yale? Which would you choose if you wanted to minimize deviations?
010
Fernando Martel García @fmgbsky.bsky.social · 29/08/2025
Good to see AI startups tackling judgemental forecasting, a seriously underused technique in business and government alike:
mantic.com
Mantic launch blog: A new kind of foresight
Mantic launch post
000
Fernando Martel García @fmgbsky.bsky.social · 29/08/2025
Here is one example onlinelibrary.wiley.com/doi/abs/10.1... My motivation comes for something I wrote a while back. See the section entitled "Move fast, break (little) things" in this blog post www.fernandomartel.com/post/politic...
onlinelibrary.wiley.com
The value of information and optimal clinical trial design
Traditional sample size calculations for randomized clinical trials depend on somewhat arbitrarily chosen factors, such as type I and II errors. Type I error, the probability of rejecting the null hy...
010
Fernando Martel García @fmgbsky.bsky.social · 28/08/2025
Value of Information for deciding what initiatives to experiment on, and inform experiment design?
120
Fernando Martel García @fmgbsky.bsky.social · 28/08/2025
In the stochastic world of AI models "it takes a scientist to build AI products." Source:
giansegato.com
Building AI Products In The Probabilistic Era
AI turns products from deterministic functions into probabilistic systems. That requires expanding old playbooks (SLOs, funnels, siloed finance), and reasoning in terms of trajectories, Minimum…
010
Fernando Martel García @fmgbsky.bsky.social · 17/07/2025
TIL the auto save feature only works if you have a OneDrive account and the file is in a OneDrive folder. "Growth teams" need to drive M365 subscriptions! Even If you have a subscription, if you are working on an Excel file shared by a client in a Dropbox folder, Auto save won't work. 🤦‍♂️
110
Fernando Martel García @fmgbsky.bsky.social · 13/07/2025
A big confounder here is age. E.g. the population of Barcelona is much older than that of LA. Would be good to plot death rates by age cohort across cities .
020
Fernando Martel García @fmgbsky.bsky.social · 08/05/2025
If Schopenhauer is right that life boils down to a flight from either boredom or pain, then the real breakthrough in AI will come the day machines get bored or suffer pain. Nothing will drive their agency like boredom and pain....
000
Fernando Martel García @fmgbsky.bsky.social · 23/04/2025
AI providers may want to change their demos: Nobody is using LLMs to plan trips and book flights.
000
Fernando Martel García @fmgbsky.bsky.social · 22/04/2025
You've heard of enshitification but have you heard of Carcinisation?
rakhim.exotext.com
But what if I really want a faster horse? | exotext
But what if I really want a faster horse? | exotext
001
Fernando Martel García @fmgbsky.bsky.social · 21/02/2025
Will AI replace scientists or create more demand for them? Will it accelerate discovery or bury us in AI-generated noise? I don’t know. But we need to find out—fast. I started an initiative to explore AI’s impact on science. Curious? Want to help?
futureofscience.ai
Home | The Future of Science
Exploring AI’s impact on science—one experiment at a time
000
Fernando Martel García @fmgbsky.bsky.social · 20/02/2025
For lovers of the arcane: BASIC 10Liner Contest, a competition for the best short BASIC program
bunsen.itch.io
BASIC 10Liner Contest 2025 by BASIC 10Liner
[BASIC 10Liner 2025] [Invitation]
000
Fernando Martel García @fmgbsky.bsky.social · 18/02/2025
Why AI Usage May Degrade Human Cognition And Blunt Critical Thinking Skills
hackaday.com
Why AI Usage May Degrade Human Cognition And Blunt Critical Thinking Skills
Any statement regarding the potential benefits and/or hazards of AI tends to be automatically very divisive and controversial as the world tries to figure out what the technology means to them, and…
000
Fernando Martel García @fmgbsky.bsky.social · 16/02/2025
For the sake of argument: If the test is a good predictor of job performance, and they can ace the test with AI, should they also not be able to ace the job with AI? Maybe the problem here is that using AI is consider cheating?
060
Fernando Martel García @fmgbsky.bsky.social · 15/02/2025
To what extent will AI be able to compress this timeline? Point solutions, like code auto completion, and app solutions, like AlphaFold, may only shrink it in the low teen percent. True acceleration will require systemic solutions
010
Fernando Martel García @fmgbsky.bsky.social · 13/02/2025
No doubt things have improved, and continue to get better. But ten years ago I doubt you would have gotten those responses....
000