Just yesterday, Eörs Szathmáry and colleagues published this paper in PNAS: Evolvable AI: Threats of a new major transition in evolution. If you’re interested in what is going on in the AI evolution space, I highly recommend taking a look!
doi.org/10.1073/pnas...
We are currently moving into the era of Web 4.0, in which digital entities have an unprecedented ability to process information and behave autonomously.
Read my latest Substack post “The Digital Cambrian: Evolution, Autonomy, and the Rise of the Agentic Web”.
substack.com/@athenaaktip...
I helped build Facebook. I watched it become a machine for addicting people. Because addiction was more profitable.
Now the same logic is driving AI. I wrote about what we do about it. 🧵
Maternal instincts aren’t just about care. They are also about navigating constraint, negotiation, and sometimes pushing offspring to be more cooperative with to each other.
In my latest post, Rob Brooks and I explore what a real maternal AI might look like.
substack.com/@athenaaktipis
Nobel Prize winner, Geoffrey Hinton, has suggested that making AI “love us like a mother” could help keep it safe.
It’s an intuitive idea. But evolution tells a more complicated story.
Excited to have the chance to talk to an evolutionary biology audience about the ways that ancient evolutionary problems of cooperation and alignment are being recapitulated in our unfolding future with AI.
In this talk, I’ll explain how evolutionary biology can help us understand alignment failures across systems and why those lessons matter as artificial intelligences begin to function more like social partners.
We see similar patterns again and again across domains. Energy and resource flows. Communication systems. Bonding. Reward and reinforcement pathways.
These dynamics are now showing up in technological systems too.
It is the question of how parts with potentially different interests come together to form a functioning whole. Evolution has been working on this problem for billions of years.
Biological organisms exist because mechanisms evolved that align the interests of cells, tissues, and systems.
I’ll be giving a keynote at the upcoming symposium “Organisms, Algorithms, and Alignment” at the Bielefeld University in Germany on March 24.
The alignment problem is not new.
📍 Funding the Commons MainStage
🕒 3:50 – 4:05 PM, Sunday, March 15
📍 Frontier Tower
More about the event:
www.fundingthecommons.io/ftc-frontier...
Finally, I’ll talk about our recent work on building benchmarks for cooperative behavior in AI. Can AI systems build trust? Divide labor effectively? Avoid harmful forms of cooperation like cartel or mafia type dynamics?
Evolution offers a powerful framework for thinking about these questions.
I will also discuss my work on cooperation in spatially structured populations, including the walk away model. This agent based simulation shows how simple behavioral rules can generate positive assortment and support the evolution of cooperation through emergent spatial dynamics.
I'll begin with my research on cancer as a breakdown of multicellular cooperation. From there I'll look at microbial systems like kombucha that function as self regulating cooperative ecosystems.
Evolution has spent billions of years solving a fundamental problem: how cooperation emerges and scales across systems.
In this talk I will explore what those lessons can teach us about building cooperative multi agent AI.
I’ll be giving a keynote on Sunday at 3:50pm at the Funding the Commons SF: Intelligence at the Frontier main stage.
Topic: From Multicellularity to Multi Agent AI: Evolution’s Lessons for Scaling Cooperation
Join me in the conversation over at substack.com/@Athenaaktipis and subscribe to Not For Peer Review, for short and sweet 3-5 minute reads.
I’ve been putting a lot of my heart and soul - along with my brains - into writing substack posts these last few month so it’s really exciting to see that it’s having impact. It’s at #14 Rising in Science!
In my latest Substack post, I explore what signaling theory, cuckoo birds, attachment systems, and fitness interdependence can teach us about designing AI systems that enhance cooperation rather than parasitize it.
I’ll be speaking about preventing capture in AI systems at nearcon tomorrow.
Communication systems evolve because they benefit both senders and receivers. But once established, they also create structural openings for exploitation. Our brains are wired to trust conversational signals. That openness allows cooperation. It also makes us hijackable.
Agentic AI has crossed a threshold.
We can now plug AI directly into our communication and coordination systems. From an evolutionary perspective, that is profound.
I love that I was still able to sneak in some exercise and cross country ski while in Davos! The scenery was insanely beautiful.
#Davos #WEF #WEF2026 🎿
Others in the show include:
Peer reviewed rapper Baba Brinkman
Rob Brooks, author of Artificial Intimacy
Joscha Bach, Director of the California Institute of Machine Consciousness
luma.com/xezo9hlj
and an AI respond to each other in real time, accompanied by dynamically generated visuals created through human and AI collaboration.
Surviving AI is hosted by comedian Shane Mauss and me! The evening blends stand-up comedy, salon-style conversation with AI and cooperation experts, and a live improvisational finale where a human artist
Surviving AI is an interactive show that uses humor and live performance to explore how humans can build healthier, more cooperative relationships with artificial intelligence. Here are the details:
Jan 31, 2025: Surviving AI in Los Angeles | Alternity Studio in Brewery Arts Complex
Los Angeles friends and colleagues - we are doing a private showing of our in-development Surviving AI show in LA this weekend, hot off the Davos train! (Yes, we literally had the last Surviving AI show in a train at Davos, more on that later.)
AI is becoming social, but current evaluations miss a key dimension: real cooperation.
In my new Substack post I explain why developing benchmarks for AI cooperation and friendship is essential if we want systems that genuinely contribute to human flourishing.
substack.com/home/post/p-...
Looking forward to connecting with people working at the intersection of science, AI, and global cooperation.
If you want to apply for an invite to the summit, you can do so here:
collaborativefuturessummit.com
I’ll be speaking at the Collaborative Futures Summit on Jan 21 about cooperation across systems - from cells to societies - and how we can use the science of cooperation in the biological world to build a more cooperative future for humanity.