Tag: Hard Problem of Consciousness

Limits for the Unlimited Mind

Mark Davis1 Comment

I’ve had a longer-than-usual hiatus from posting due to adding a new living arrangement to our small collection. Acquiring and moving furniture, appliances, toiletries, bedding, audio equipment, and cleaning supplies between spaces has an almost unlimited ability to distract. And when a quiet final settles in, there remains a restlessness about optimizing experiences and priorities that takes time to psychically draw down. It’s the clattering opposite of focusmaxxing. On the positive, new places constantly reset and tune expectations about what is important and what matters little in a life, with people, ideas, and new experiences always rising above the glimmer of these damn possessions.

This rich inner life, this sense of a self that powers the great religions and mystical traditions of the world, this atma and soul, these ideas of impermanence, justice, and wonder—can they all be a product of a versatile connective matrix that computes these optimizing decisions while simultaneously creating a vital subjective sense of identity? There is a theory called unlimited associational learning (UAL) that posits exactly that. Our adaptation to the world is one of temporal superiority. A plan and a reaction for a conscience entity is a function for pulling time apart and predicting the future, from the cause-effect of seasonal migrations of herd animals to estimating the supportive and defensive capabilities of mates.

The essential hinge is that this consciousness, this sense of being someone, must on average be more effective than the other trajectories through the phylogenetic, epigenetic, and ontogenetic state spaces. And that’s even with all the flaws like misperceptions, biases, neurological injuries, and schizophrenia. Proving this with any sort of mathematical precision has an almost impossible bar to meet since it would require that a model first is shown to be conscious with the set of UAL-sufficient hallmarks, like global accessibility, integration over time, agency and embodiment, and so on.… Read the rest

Sentience is Physical

Mark DavisLeave a comment

Sentience is all the rage these days. With large language models (LLMs) based on deep learning neural networks, question-answering behavior of these systems takes on curious approximations to talking with a smart person. Recently a member of Google’s AI team was fired after declaring one of their systems sentient. His offense? Violating public disclosure rules. I and many others who have a firm understanding of how these systems work—by predicting next words from previous productions crossed with the question token stream—are quick to dismiss the claims of sentience. But what does sentience really amount to and how can we determine if a machine becomes sentient?

Note that there are those who differentiate sentience (able to have feelings), from sapience (able to have thoughts), and consciousness (some private, subjective phenomenal sense of self). I am willing to blend them together a bit since the topic here isn’t narrowly trying to address the ethics of animal treatment, for example, where the distinction can be useful.

First we have the “imitation game” Turing test-style approach to the question of how we might ever determine if a machine becomes sentient. If a remote machine can fool a human into believing it is a person, it must be as intelligent as a person and therefore sentient like we presume of people. But this is a limited goal line. If the interaction is only over a limited domain like solving your cable internet installation problems, we don’t think of that as a sentient machine. Even against a larger domain of open-ended question and answering, if the human doesn’t hit upon a revealing kind of error that a machine might make that a human would not, we remain unconvinced that the target is sentient.… Read the rest

Wordle and the Hard Problem of Philosophy

Mark Davis2 Comments

I occasionally do Wordles at the New York Times. If you are not familiar, the game is very simple. You have six chances to guess a five-letter word. When you make a guess, letters that are in the correct position turn green. Letters that are in the word but in the wrong position turn yellow. The mental process for solving them is best optimized by choosing a word initially that has high-frequency English letters, like “notes,” and then proceeding from there. At some point in the guessing process, one is confronted with anchoring known letters and trying to remember words that might fit the sequence. There is a handy virtual keyboard displayed below the word matrix that shows you the letters in black, yellow, green, and gray that you have tried, that are required, that are fit to position, and that remain untested, respectively. After a bit, you start to apply little algorithms and exclusionary rules to the process: What if I anchor an S at the beginning? There are no five-letter words that end in “yi” in English, etc. There is a feeling of working through these mental strategies and even a feeling of green and yellow as signposts along the way.

I decided this morning to write the simplest one-line Wordle helper I could and solved the puzzle in two guesses:

Sorry for the spoiler if you haven’t gotten to it yet! Here’s what I needed to do the job: a five letter word list for English and a word frequency list for English. I could have derived the first from the second but found the first first, here. The second required I log into Kaggle to get a good CSV searchable list.… Read the rest