Marcelo Novaes (@marnovo) header
Marcelo Novaes (@marnovo) profile picture

≔⠀VP Strat & Ops @linuxfoundation + COO & Co-founder @athenian ⠀∵⠀ ∏⠀@sourcedtech, Ronit, fisherVB, GS, S&P, IE, USP et al.⠀∴⠀ ∞⠀mentī quōlibet īre licet⠀∎

Vault 0 ⇌ BR ⇌ ES ⇌ US
Joined October 2008
Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)Photo by Marcelo Novaes (@marnovo)
“Ars longa, vita brevis, occasio praeceps, experimentum periculosum, iudicium difficile.” –Hippokrátēs
1
1
2
16
Marcelo Novaes retweeted
New Paper! Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents A longstanding goal of AI research has been the creation of AI that can learn indefinitely. One path toward that goal is an AI that improves itself by rewriting its own code, including any code responsible for learning. That idea, known as a Gödel Machine, proposed by @SchmidhuberAI over two decades ago, is a hypothetical self-improving AI. It optimally solves problems by recursively rewriting its own code when it can mathematically prove a better strategy, making it a key concept in meta-learning or “learning to learn.” While the theoretical Gödel Machine promised provably beneficial self-modifications, its realization relied on an impractical assumption: that the AI could mathematically prove that a proposed change in its own code would yield a net improvement before adopting it. Sakana AI, in collaboration with Jeff Clune’s lab at UBC, proposes something more feasible: a system that harnesses the principles of open-ended algorithms like Darwinian evolution to search for improvements that empirically improve performance. We call the result the Darwin Gödel Machine. DGMs leverage foundation models to propose code improvements, and use recent innovations in open-ended algorithms to search for a growing library of diverse, high-quality AI agents. Applied to practical tasks, we implemented Darwin Gödel Machine as a self-improving coding agent that rewrites its own code to improve performance on programming tasks. It creates various self-improvements, such as a patch validation step, better file viewing, enhanced editing tools, generating and ranking multiple solutions to choose the best one, and adding a history of what has been tried before (and why it failed) when making new changes (see the attached video). We believe that Darwin Gödel Machines represent a concrete step towards AI systems that can autonomously gather their own stepping stones to learn and innovate forever!
36
199
24
1,031
105,147
Here's a fun AI story: a security researcher noticed that large companies' AI-authored source-code repeatedly referenced a nonexistent library (an AI "hallucination"), so he created a (defanged) malicious library with that name and uploaded it. 1/
57
1,362
252
5,849
1,178,962
"Done with the drama, folks. Let's get back to work. And use Teams."
We remain committed to our partnership with OpenAI and have confidence in our product roadmap, our ability to continue to innovate with everything we announced at Microsoft Ignite, and in continuing to support our customers and partners. We look forward to getting to know Emmett Shear and OAI's new leadership team and working with them. And we’re extremely excited to share the news that Sam Altman and Greg Brockman, together with colleagues, will be joining Microsoft to lead a new advanced AI research team. We look forward to moving quickly to provide them with the resources needed for their success.
1
3
259
Marcelo Novaes retweeted
“You won’t appreciate the wide open time to think and explore that you had in your 20s until you completely are scheduled within an inch of your life in your 40s.” Message in a bottle I wish I could send to myself
55
243
32
2,673
326,406
Marcelo Novaes retweeted
Will share more soon but this quote from @Redpoint’s post is a great tl dr; “Poolside aims to unlock humanity’s potential by pursuing AGI for software creation, with the fundamental belief that the transitional path for humanity to AGI is by building for specific capabilities instead of a general purpose approach. For Poolside, that starts with unlocking the power of software creation and understanding and building the systems that will create all future systems.” medium.com/redpoint-ventures…
Couldn’t be more excited about what @jasoncwarner and @eisokant are building. Check out why @ericabrescia and I are thrilled to get to partner with the perfect team to take on a world-changing vision to power the future of software creation: medium.com/redpoint-ventures…
9
6
1
65
33,443
Marcelo Novaes retweeted
People are testing large language models (LLMs) on their "cognitive" abilities - theory of mind, causality, syllogistic reasoning, etc. Many (most?) of these evaluations are deeply flawed. To evaluate LLMs effectively, we need some principles from experimental psychology.🧵
36
459
78
1,823
549,157
. @Wolfram_Alpha as the way to bring computational knowledge superpowers to #ChatGPT: writings.stephenwolfram.com/…
166
Marcelo Novaes retweeted
GPT3 is powerful but blind. The future of Foundation Models will be embodied agents that proactively take actions, endlessly explore the world, and continuously self-improve. What does it take? In our NeurIPS Outstanding Paper “MineDojo”, we provide a blueprint for this future:🧵
48
492
92
2,429
After being knee-deep & closely watching the AI/ML renaissance the last decade, @openAI #ChatGPT is a clear revolutionary step in usefulness and user experience. Even recent DALL-E and StableDiffusion feel like us struggling to predict the next coherent tokens just years ago...
1
2
Enough talk, just try it on your own and be mesmerized: chat.openai.com/chat You will want to have this with you every day, and the "google search, read, put it together, refine/repurpose" flow that fueled our collective brain will often feel antiquated.
1
1
Bonus 1: Needless to say that interacting with the likes of "AI assistants" as Google Assistant, Siri, Alexa, Cortana will also feel even dumber... Bonus 2: imagine what connecting these meta-models to your accounts across your software/data will mean as a "second brain".
Marcelo Novaes retweeted
I have spent almost twenty years of my life looking at China's balance of payments data in one way or another. Right now neither the current account numbers nor the financial account numbers quite fit together. 1/x cfr.org/article/chinas-econo…
11
126
15
486
Marcelo Novaes retweeted
A Brief History of Nobody Wants to Work Anymore 🧵
2,437
72,409
12,161
261,368
If these resonate: –Software is eating the world –Talent wins games, teamwork & intelligence win championships –Getting 1% better every day counts for a lot in the long-run –There's nothing so useless as doing efficiently that which should not be done at all Then check @athenian
We are delighted to announce that we raised a ✨$6M✨ seed round, led by @PointNineCap, with participation from @frst_vc, @Xavier75, @20vcFund, @taylor_w_c, and @airstreet alongside stellar angels! 🚀
1
1
Software + engineering + leadership + alignment + teamwork + change + effectiveness + improvement + efficiency + data = @athenian Find out how data-enabled engineering is the future of software development: –Blog: athenian.com/blog/the-future… –TechCrunch: techcrunch.com/2022/03/02/at…
1
Marcelo Novaes retweeted
The ongoing consolidation in AI is incredible. Thread: ➡️ When I started ~decade ago vision, speech, natural language, reinforcement learning, etc. were completely separate; You couldn't read papers across areas - the approaches were completely different, often not even ML based.
298
1,584
251
7,863
Every interesting outcome in my life was unpredictable and was the result of good habits, strictly practiced. This is the opposite of “vision” “big, hairy, audacious goals” and “accomplishments.” The best kind of success is continuous, every day, not occasional.
81
533
118
4,177