FAIRC

Manifesto

FAIRC · Manifesto2026

On Understanding, with an Application to Artificial Intelligence

Manifesto

1A redefinition

FAIRC is born out of a redefinition of Artificial Intelligence research. Our mission is to pursue AI with the curiosity and rigor of the sciences that came before it: seeking understanding before application, and explanation before optimization.

Richard Feynman1Physicist, who shared the 1965 Nobel Prize in Physics with Julian Schwinger and Sin-Itiro Tomonaga for their work on quantum electrodynamics. The remark was made in a 1981 interview for the BBC. was once asked what winning the Nobel Prize meant to him, and his answer brushed the honor aside: what mattered, he said, was “the pleasure of finding the thing out” [1]; the thrill of discovery itself, and knowing the work would go on to be useful to someone else.

That is the spirit we build from.

2Dissatisfaction

We are dissatisfied with how little we truly understand about intelligence, with a culture that too often rewards uniformity, and with an AI landscape still dominated by brute force: more data, more compute, more scale.2The case for scale has well-known statements, among them Sutton’s essay [2] and the empirical scaling laws of Kaplan et al. [3]. We take them seriously; they are part of what we want explained. AI has advanced rapidly, yet so much of it remains centered around applications, while fundamental questions remain untouched. We still understand remarkably little about how intelligence can emerge from limited experience,3Figure 1 on our front page is a small, live instance of this question. or what mechanisms allow intelligent systems to learn, generalize, and reason.

Too often, we start from applications, from what we need, rather than from understanding what we have and questioning it.

Thus is born our vision

To answer questions no one is answering, and to ask questions no one is asking.

3The rules beneath the moves

We believe that mathematics, philosophy, and our understanding of reality lay the groundwork for everything anyone has built. The science of Artificial Intelligence is no exception. Beneath every complexity and capability of AI lies an elegant interplay of mathematics and intuition. Our approach is not to stand as a spectator watching the player improve, but to understand the game itself: the rules beneath the moves.

Rule 110011011102

Known rules
First row

Hover an entry to see every cell it wrote. Press it to flip its output and run the world again.

Figure 1. The rules beneath the moves. An elementary cellular automaton [4]: every cell in a row takes its state from the three cells above it, by the eight-entry rule shown, and time runs downward. Reading the eight outputs as a binary number names the rule. Rule 30 produces apparent randomness, rule 90 draws a Sierpiński triangle, and rule 110 is capable of universal computation [5]. Everything in the diagram is decided by those eight bits.

4Frontiers

At FAIRC, we pursue multiple frontiers in foundational Artificial Intelligence, including Language Modelling, Neuroscience-Inspired Intelligence, Knowledge Repre­sen­ta­tion, and Rein­force­ment Learning.4Each frontier is drawn as a live plate on our front page. Our scope is broad by design. We pursue any line of inquiry that advances the mechanistic understanding of intelligent systems: from the mathematics of representation to the biology of cognition, and from the dynamics of learning to embodied agents.

We do not pursue questions because they are fashionable, scalable, or immediately useful. We pursue them because they remain unanswered.

Dissatisfaction is the thrust; discovery is what answers it.

The Frontier AI Research Collective

Citation

@misc{fairc2026manifesto,
  title        = {On Understanding, with an Application
                  to Artificial Intelligence},
  author       = {{Frontier AI Research Collective}},
  year         = {2026},
  howpublished = {\url{https://fairc.org/manifesto}},
  note         = {Manifesto}
}
BibTeX

References

  1. [1]Feynman, R. P. (1981). The Pleasure of Finding Things Out. Horizon, BBC Television.
  2. [2]Sutton, R. S. (2019). The bitter lesson. Incomplete Ideas.
  3. [3]Kaplan, J., McCandlish, S., Henighan, T., Brown, T. B., et al. (2020). Scaling laws for neural language models. arXiv:2001.08361.
  4. [4]Wolfram, S. (1983). Statistical mechanics of cellular automata. Reviews of Modern Physics, 55(3), 601–644.
  5. [5]Cook, M. (2004). Universality in elementary cellular automata. Complex Systems, 15(1), 1–40.