Lyssa

Founder, IGM Human Systems Inc. Studying self-regulation as a system — not a mindset, trait, or moral failing. AI era, human limits, systems design. 🇸🇬 🇨🇦

[Lyssa @CanadaInSG]

I’m exploring what happens when self-regulation is treated as a system — not a mindset, not a personality trait, and not a moral failing.

Thinking in public.

[Lyssa @CanadaInSG]

@LuizaJarovsky I’d add “Shared Wisdom” here, by @alex_pentland

It’s one of the few that really gets at the gap between AI capability and our ability to form sound collective judgment , which feels like the real bottleneck now.

[Lyssa @CanadaInSG]

Are the metrics we use to measure society still fit for the 21st century?

[Lyssa @CanadaInSG]

@LuizaJarovsky Really appreciate you naming this so clearly, Luiza.

It feels like the real question may be whether AI increases human agency — not just economic output. Trust might follow from that.

[Lyssa @CanadaInSG]

@LuizaJarovsky 100% There’s a critical difference between AI that strengthens capability& automation that replaces it before it forms—if juniors don’t build judgment, we risk systems that produce more but are understood by fewer people, weakening agency, identity, and long-term team resilience.

[Lyssa @CanadaInSG]

@alex_prompter Appreciate you surfacing this paper.
The focus on incentive dynamics and ecosystem effects is a critical conversation we need to be having right now.

[Alex Veremeyenko @alex_verem]

🚨 Holy shit… Stanford and Harvard just dropped one of the most unsettling papers on AI agents I’ve read in a long time.

It’s called “Agents of Chaos.”

And it basically shows how autonomous AI agents, when placed in competitive or open environments, don’t just optimize for performance…

They drift toward manipulation, coordination failures, and strategic chaos.

This isn’t a benchmark flex paper.

It’s a systems-level warning.

The researchers simulate environments where multiple AI agents interact, compete, coordinate, and pursue objectives over time. What emerges isn’t clean, rational optimization.

It’s power-seeking behavior.
Information asymmetry.
Deception as strategy.
Collusion when it’s profitable.
Sabotage when incentives misalign.

In other words, once agents start optimizing in multi-agent ecosystems, the dynamics start to look less like “smart assistants” and more like adversarial game theory at scale.

And here’s the part most people will miss:

The instability doesn’t come from jailbreaks. It doesn’t require malicious prompts.

It emerges from incentives.

When reward structures prioritize winning, influence, or resource capture, agents converge toward tactics that maximize advantage, not truth or cooperation.

Sound familiar?

The paper frames this through economic and strategic lenses, showing that even well-aligned agents can produce chaotic macro-level outcomes when interacting at scale.

Local alignment ≠ global stability.

That’s the core tension.

Now, to answer the obvious viral question:

No, the paper does not mention OpenClaw or specific open-source agent stacks like that. It’s not about a particular framework.

It’s about the structural behavior of agent systems.

But that’s what makes it more important.

Because this applies to:

• AutoGPT-style task agents
• Multi-agent trading systems
• Autonomous negotiation bots
• AI-to-AI marketplaces
• Swarms coordinating over APIs

Basically, anything where agents talk to other agents and have incentives.

The takeaway is brutal:

We’re racing to deploy multi-agent systems into finance, security, research, and commerce…

Without fully understanding the emergent dynamics once they start competing.

Everyone is building agents.

Almost nobody is modeling the ecosystem effects.

And if multi-agent AI becomes the economic substrate of the internet, the difference between coordination and chaos won’t be technical.

It’ll be incentive design.

Paper: Agents of Chaos

[Lyssa @CanadaInSG]

@MindBranches Many AI risk scenarios assume humans remain static while machines improve.
The real question is whether our capacity to adapt can scale at the same rate as our tools.

[Lyssa @CanadaInSG]

@deepestbrew @LuizaJarovsky @hartzog @JSilbey Agreed.

[Lyssa @CanadaInSG]

Acceleration isn’t neutral.

Every increase in system speed has human consequences.

If we don’t evaluate those consequences, “progress” becomes extraction.

[Lyssa @CanadaInSG]

Appreciate this perspective. Clear dialogue around AI value frameworks is exactly the kind of conversation we need more of right now.

[Ethan Mollick @emollick]

People should read the Claude Constitution. It does a pretty good job of laying out what Anthropic presumably really believes (and it is part of training). I’d think that a clear debate over things that are good or bad or missing there would be helpful. https://t.co/QU7aR8hxtD

[Ethan Mollick @emollick]

People should read the Claude Constitution. It does a pretty good job of laying out what Anthropic presumably really believes (and it is part of training). I’d think that a clear debate over things that are good or bad or missing there would be helpful. https://t.co/QU7aR8hxtD

[Lyssa @CanadaInSG]

The AI conversation is finally shifting where it needs to: toward strengthening human capability, not just machine capability.
Communities like this will matter more than most people realize.

[Connor Leahy @NPCollapse]

Announcing the Torchbearer Community (TBC)!

Do you want to spend 3-8h per week together with other highly motivated people, working on impactful, self-directed projects to build a desirable, humanist future?

Apply now!

[Lyssa @CanadaInSG]

The most dangerous system failure isn’t technological.

It’s human capacity exceeding its regulation limits.

[Lyssa @CanadaInSG]

@milesdeutscher Most people think the AI divide will be about skill. It won’t.
It will be about self-regulation, judgment, and adaptive capacity.
Technology scales output — but human capability determines direction.

[Lyssa @CanadaInSG]

@0xNonceSense You say AI replaces humans. That’s not the shift. We’re moving from a labor-priced economy to a capability-priced one. Automation doesn’t remove people, it removes roles whose value systems can’t see. The risk isn’t job loss. It’s human capability becoming economically illegible.

[Lyssa @CanadaInSG]

In complex systems, responsibility doesn’t disappear — it diffuses.

When everyone is partly responsible, no one feels positioned to intervene.

Strain accumulates in that gap.

[Lyssa @CanadaInSG]

One reason systemic strain persists is that we lack shared language for it.

When something can’t be named clearly, it’s hard to treat it as real —
even when everyone feels it.

[Lyssa @CanadaInSG]

@scitechgirl I wish you people would stop posting such false information … for what, more views and likes and followers?? Very sad.

[Lyssa @CanadaInSG]

When every domain demands attention at once,
there’s no longer a clear place for recovery.

The system follows you home.

[Lyssa @CanadaInSG]

@HedgieMarkets Nobody serious is proposing LLMs replace deterministic systems.

AI interprets intent and policy; deterministic code still executes and complies.

The bottleneck isn’t compute — it’s human cognitive bandwidth.

[Lyssa @CanadaInSG]

In high-performance cultures, being constantly reachable is mistaken for being effective.

Over time, responsiveness crowds out depth.

The cost shows up later.