# Nils Baierl — the capability frontier

> AI researcher · Multi-agent systems · Claim verification · Fraunhofer AISEC · OTH Amberg-Weiden

## About

I'm an AI researcher based in Bavaria, Germany. My work sits at the intersection of multi-agent systems, cognition, and information integrity.

Focus areas: intelligent agents, claim verification, misinformation detection, AI alignment through deployment. Currently at Fraunhofer AISEC, working on agent systems for information integrity.

Available for: AI automation projects, agent system design, research collaborations, consulting.

Contact: nils.baierl1@gmail.com

---

## CV — Nils Baierl

### Education

- **2022–2026** BSc Artificial Intelligence, OTH Amberg-Weiden. Grade: 1.6. Thesis: "The Significance of Artificial Intelligence for the Realization of Transhumanist Visions" — Grade: 1.0. 210 ECTS.
- **2019–2020** Computer Science (2 semesters), University of Bayreuth
- **2010–2019** Gymnasium Eschenbach — Abitur

### Research & Work

- **2024–2026** Research Intern, Fraunhofer AISEC — Cognitive Security Technologies, Garching/Munich. LLM-based multi-agent systems for automated fact-checking and claim verification.
- **2023–2024** Student Software Developer, NeuroForge Solutions GmbH. Integration of large language models into web applications.
- **2020–2022** Creative Work — Music & Video Production. Full-time music production (Logic Pro X) and video production (Adobe Premiere Pro).

### Technical Skills

- **Languages:** Python, TypeScript, Java, C++
- **Frameworks:** vLLM, Hugging Face Transformers, OpenClaw, AutoGen, PyTorch, Keras, Scikit-learn, FastAPI, Docker
- **Areas:** NLP, generative AI, agentic systems, multi-agent orchestration, computer vision

### Languages

- German: Native
- English: C1

---

## Essays

### Nobody Planned This — Artificial Superintelligence as an Emergent Phenomenon

**Abstract:** This essay argues that ASI will not emerge through parameter scaling alone, but through the emergent properties of multi-agent systems. Drawing on Minsky's Society of Mind, Langton's Ant, and recent empirical evidence from OpenAI (Baker et al. 2019) and ICLR 2026 (Riedl), the core hypothesis is that superintelligence arises from the scaling of many agents in interaction — topology scaling — rather than from monolithic model scaling.

Key ideas:
- Minsky (1986): Intelligence emerges from the interaction of many simple, unintelligent agents
- Langton's Ant: Emergent order from simple local rules
- Baker et al. 2019: Multi-agent hide-and-seek produced six emergent strategies (shelter-building, ramp use, box surfing)
- Riedl (ICLR 2026): Multi-agent LLM systems show measurable, steerable emergence in coordination tasks
- Hypothesis: ASI arises from topology scaling (agent interactions), not parameter scaling

Full essay (PDF) available at: /writing/nobody-planned-this-full.pdf

### Is Unregulated AI Research a Good Idea?

**Abstract:** An exploration of dual-use AI research risks, alignment and control concerns, and the asymmetry between regulating research vs. deployment. Argues that while deployment regulation is feasible and desirable, regulating knowledge production faces fundamental challenges — global jurisdiction gaps, bureaucratic capture, innovation slowdown — that make professional norms and ethical responsibility within the research community the most practical lever.

---

## Social

- LinkedIn: https://de.linkedin.com/in/nils-baierl-023584325
- GitHub: https://github.com/Cachet23