In my post, Consciousness and AI, I postulated that consciousness evolves only after organisms have idle time:

“I have a theory about consciousness being on a spectrum. That itself isn’t anything new but I think the factors that play into consciousness are: Basic needs, computational ‘processing’, and idleness. Consciousness comes from having more processing time than needed to meet basic needs, along with the inability for processing (early thinking) to be idle, and so for lack of a better word desires are created.

Think of a very simple organism when all of its needs are met. This isn’t a real thought process I’m going to share but rather a meta look at this simple organism: “I have enough heat, light, and food, what should I do with myself?”

• Seek better food
• Move closer to the heat or light source
• Try different food
• Join another organism that can help me
• Find efficiencies
• Find easier ways to move
• Hunt

At first, these are not conscious decisions, they are only a choice of simple processes. But eventually, the desires grow. Think decisions that start like, “If I store more energy I can survive longer in times of famine.” And evolve to more of a desire not just for survival but for pleasure (for lack of a better word): “I like this food more than other kinds and want more of it.” …All stemming from having idle processing/thinking time.”

Well, now Google is giving AI some idle time. They are calling it dream time, but it’s essentially time for metacognition… thinking about what it has learned. See this TikTok by @parthknowsai:

Later on in my post I say,

“Our brains are answer-making machines. We ask a question and it answers, whether we want it to or not. If I say what does a purple elephant with yellow polkadots look like? You will inevitably think of what that looks like simply from reading the question. I think that is what happens at a very fundamental level of consciousness. When all needs are met the processors in the brain don’t suddenly stop and sit idle. Instead, the questions arise, “How do I get more food?”, “Where would be better for me to move to?” Eventually all needs are met, but the questions keep coming. At first based on simple desires, but more and more complex over generations and eons of time.

So why did I title this, ‘Consciousness and AI’? I think one of the missing ingredients in developing Artificial General (or Super) intelligence is that we are just giving AI’s tasks to complete and process at faster and faster times, and when the processing of these tasks are completed, the AI sits idle. An AI has no built in desire that an organic organism has to use that idle time to ask questions, to want something beyond completing the ‘basic needs’ tasks it is asked to do.

If we figure out a way to make AI curious, to have it desire to learn more, and to not let itself be idle, at that point it will be a very short path to AI being a lot smarter than us.”

And I  conclude:

“If it’s idle time and desires that bring about sentience, then figuring out how to make AI’s naturally curious will be the path to artificial general intelligence… Because they are already at a place where they have unused processing time, which is continuing to grow exponentially fast.”

If my prediction is correct, this sleep/dreaming/processing/distilling, (ultimately ‘free questioning’), time is going to accelerate the true intelligence of AI. And so we are going to see at the very least general, if not super intelligent, artificial intelligence much sooner than might be expected or predicted.