Professeur à @Paris_Dauphine #IA #management #Société #Culture
Paris
Joined March 2009
- Tweets48.2K
- Following6.2K
- Followers7.3K
- Likes15.2K
Pinned Tweet
#backtoaFreeJazz a sensible selecta #Jazz #Funk #soul #galactic #free a 80 years journey #blackmusic soundcloud.com/user-78356425…
@benavent retweeted
TaskNet is live!
By accurately mapping the tasks workers perform at their jobs, the project will help us gain a better understanding of AI's impact on work.
Explore Tasknet here: tasknet.stanford.edu
si #Parcoursup doit mourrir (mais je crains que ce soit un #phenix), il faut dans le même mouvement bruler #adum (même si avec un peu d'#UX et de respect de la vie personnelle on pourrait le ressusciter). les #doctorants me comprendrons, j'espère. 🙃
@benavent retweeted
Les premiers rassemblements ont commencé à se former à la Porte d'Italie à Paris en ce début d'après-midi du 24 octobre.
Ce point de départ correspond au parcours officiel de substitution arrêté par la Préfecture de police vers la Porte de Vincennes.
Un important dispositif des forces de l'ordre entoure déjà le secteur pour filtrer les accès et sécuriser les boulevards.
teapawt.org/maiza6393/status/21139…
Source : @maiza6393 @PAT_SB79
Médias @PAT_SB79
@benavent retweeted
Google just open-sourced Harness².
(turn any harness into a self-improving system)
a harness is everything wrapped around the model, like the system prompt, tools, memory, skills, and subagents. today a person designs it once, and the agent runs every new task with that same fixed setup.
Harness² lets the agent refine that setup for each new task on its own, with the model weights left untouched.
on every task, the agent proposes edits to its harness, runs a few candidate versions, and composes a better harness by comparing how those runs behaved.
because it learns from the differences between runs, it needs no correct answer or grader for the task.
that covers a single task. Harness² also carries what it learns from one task into the next.
after each task, it saves the harness it settled on along with the lessons that held up across the runs. the next time it refines a harness, it starts from those notes instead of from scratch.
this means every finished task makes the next round of edits more informed.
Google tested this with Gemini 3.5 Flash and GLM-5 on legal, office, and professional work benchmarks. one round of edits was enough to beat their original harnesses, and giving the agent more rounds pushed scores higher.
those extra rounds can go wide or deep. the agent either tries several candidate harnesses at once or keeps refining a single one step by step, using OpenCode or Codex to carry out the work.
you can find the Harness² repo here: github.com/google-research/h…
I wrote about a similar idea for agents already running in production, where the harness should repair itself after a failure instead of waiting for someone to debug the trace by hand.
that piece walks through how @Cometml's Opik (open-source) closes this loop, going from a failing trace to a proposed fix to a regression test that stops the same failure from coming back. it's quoted below.
Maintenant, il faut bien comprendre que lorsqu'on écrit c'est pour des #humains et pour des #machines, et que les machines sont plus utilisées par d'autres machines que par les humains. Il va falloir apprendre à vivre de manière #symbiotique #AI
AI agents are becoming the main user of company software, with humans approving its actions instead of entering data themselves.
So it should be metered accordingly.
ERP[.]AI has released a platform for business software that charges $0 per seat and bills the machines instead.
Seats, apps, roles, permissions and manual record work in its interface cost nothing, while API-key access, imports, workflows, agents and AI draw down prepaid credits.
Those credits come in one-time $5, $20 and $100 packs, and every new account gets $100 of starting credit that lasts 1 month.
The meter is fine-grained, because $1 buys 4,000 record writes, 2,000 workflow runs or 2,000 agent executions, and AI is billed at the model’s exact cost plus an 8% agent fee.
Most of that metered work comes from Proto, a desktop agent that sets up data, approvals, permissions and dashboards in ERP[.]AI from a plain description of how a business runs.
@benavent retweeted
AI agents are becoming the main user of company software, with humans approving its actions instead of entering data themselves.
So it should be metered accordingly.
ERP[.]AI has released a platform for business software that charges $0 per seat and bills the machines instead.
Seats, apps, roles, permissions and manual record work in its interface cost nothing, while API-key access, imports, workflows, agents and AI draw down prepaid credits.
Those credits come in one-time $5, $20 and $100 packs, and every new account gets $100 of starting credit that lasts 1 month.
The meter is fine-grained, because $1 buys 4,000 record writes, 2,000 workflow runs or 2,000 agent executions, and AI is billed at the model’s exact cost plus an 8% agent fee.
Most of that metered work comes from Proto, a desktop agent that sets up data, approvals, permissions and dashboards in ERP[.]AI from a plain description of how a business runs.
🤝 Paid partnership
For 25 years, Salesforce, SAP and enterprise software were built around one assumption: humans operate the software.
We’re changing that.
Introducing ERP•AI and Proto: business software built for agents, and an agent that operates business software.
Describe the system you need. ERP•AI builds it.
Then Proto takes over. It manages records, runs workflows and executes actions inside the software you just created.
This is Headless SaaS.
1: get ERP•AI
2: build your SaaS
3: post below
Best one wins a MacBook. First 100 decent entries get $100.
(a message from Steve Papa / founding team)
@benavent retweeted
Today @GoogleResearch is launching Deep Review — an agentic system designed to stress-test manuscripts, design docs, and complex proposals before submission or publication. After pilot partnerships analyzing over 15,000 research papers, we’re making Deep Review globally accessible to help researchers, engineers, and professionals bring structured critique to their daily workflows.
There is a huge opportunity for specialized personal agents.
Here is a great example:
@OdessiaTravel just launched one for travel and experiences. You text it what you want, and it plans and books the trip, the table, or the tickets.
It also keeps watching after you ask. You can set daily watches on tables, shows, and fares, and it pings you when something opens up.
The thing is, like specialized and general models, I think we will see the same for personal agents.
Love seeing more of these pop up recently.
@benavent retweeted
There is a huge opportunity for specialized personal agents.
Here is a great example:
@OdessiaTravel just launched one for travel and experiences. You text it what you want, and it plans and books the trip, the table, or the tickets.
It also keeps watching after you ask. You can set daily watches on tables, shows, and fares, and it pings you when something opens up.
The thing is, like specialized and general models, I think we will see the same for personal agents.
Love seeing more of these pop up recently.
🤝 Paid partnership
@benavent retweeted
🧵1/10 Given the recent reactions of mathematicians over rapid advances in AI, I think it is important to remember that technological change has historically been a very personal event for people.
Over the history of technological progress, looms have fallen silent, switchboards have gone dark, and factory lines have shrunk bringing with them a human crisis of identity and what it means to be a master of one's craft.
@benavent retweeted
Extreme poverty ending 👇
China reduced extreme poverty fast, but Indonesia hasn’t been far behind.
China is often the poster child for rapid reductions in poverty, and for good reason. In the early 1980s, over 90% of its population lived in extreme poverty, but by the early 2020s, that number had dropped to nearly zero.
Some might assume that China is the only reason why global extreme poverty has declined. But that’s wrong: many countries have seen dramatic reductions in poverty, too. Indonesia is one clear example; it’s shown alongside China on the chart.
In 1984, 86% of Indonesians lived on less than $3 per day. By 2025, this had fallen to less than 4%. While it didn’t quite match China’s decline, it has still been impressive. The number of people living in extreme poverty has fallen from 140 million to about 10 million.
Note that the international poverty line is extremely low, defined as people living on less than $3 per day. But Indonesia has also made progress when measuring higher poverty lines.
(written by @_HannahRitchie and @parriagadap)
@benavent retweeted
I've had so many insightful conversations over the years with @lessig, an expert on AI regulation and legislation, including more recent ones on the cyber incidents of this past summer.
Happy to share our thoughts in this new @TheEconomist op-ed, where we outline concrete actions Congress could consider to protect the public from AI’s major risks:
economist.com/by-invitation/…
@benavent retweeted
Political scientists and economists: we're looking for researchers with deep expertise to come work with us for a unique 4-month fellowship at @AnthropicAI. You'll work closely with us to evaluate, improve, and apply Claude's capabilities in your field. It's going to be extremely fun and interesting work, and definitely worth taking a short leave from your university for if you need to!
More info here:
job-boards.greenhouse.io/ant…
