C. elegans vs. human-level AI

Reading the Wikipedia entry on Caenorhabditis elegans and how much we already understand about this small organism and its 302 neurons makes me even more skeptical of the claim that a human-level artificial intelligence (short: AI) will be created within this century.

C. elegans

C. elegans, by Bob Goldstein

Its pattern of connectivity, or “connectome”, has been completely mapped and we have the computational capacity to simulate it. Yet nobody is able to do so. Its genome is completely sequenced.

Many different people and teams of people have been studying this little nematode for decades. Yet, as John Baez once formulated it, nobody is able to create an AI hat could navigate autonomously in a real-world environment and survive real-world threats and attacks with approximately the skill of C. elegans.

Would it be wrong to take this as evidence against human-level AI? If so, how? What makes you believe that it will be possible to create a human-level AI from scratch before it is possible to copy the skills of an already existing organism that is qualitatively and quantitatively many orders of magnitude less intelligent than humans?

Further reading:


  • Pingback: Alexander Kruel · MIRI/LessWrong Critiques: Index()

  • Davide

    there are several pitfalls in your argumentation.

    First of all, AI doesn’t mean neuron simulation. I don’t see the relation between simulating a worm brain with AI (that is more Computational neuroscience).

    the goal of AI is not to replicate one by one the neurons in the brain.

    Even though some biological concepts can inspire AI algorithms, this doesn’t mean that we need to simulate a brain.
    The definition of AI is also tricky. The worm you mention is not intelligent at all, it has probably some innate behaviours, more related to control than intelligence.
    I’m sure you can have some robot (program) that shows similar skills. Actually you can do better, that is learn from experience how to move (e.g., reinforcement learning), instead the worm has probably pre-configured skills, since it has a really simple brain.
    you also talk about create “human-level AI from scratch”, why from scratch? you can start from the current knowledge on AI.

    maybe we will never be able to perfectly simulate a worm brain but this have nothing to do with AI. BTW I know that it is already possible simulate a brain of the size of a rat.

  • Pingback: Alexander Kruel · Quick review of RobBB’s ‘Engaging Introductions to AI Risk’()

  • Guest

    Are you familiar with OpenWorm? http://www.openworm.org/