artificial general intelligence will become a commodity.
personal intelligence (PI) – the intelligence that represents an individual's unique intuition, judgement, and taste – will be scarce and valuable.
people will own and train their personal intelligences to enable them to scale themselves to others.
for example, prompting Akira Kurosawa's PI to generate a video will apply his unique taste and judgement to create the output. similarly, asking Andrej Karpathy's PI to explain a complex topic will simplify it in the way he does.
in the way that we currently choose AGI models, we will also choose PIs that our prompt passes through. the same prompt that passes through the same AGI model, but different PIs, will produce very different outputs – "colored" by a combination of the PIs' unique tastes.
the more advanced general intelligence becomes, the more depth personal intelligence can represent. it will take some time before we can create PIs that can represent even 5% of an individual's unique intuition, judgement, and taste. i don't think our technology is there yet (could be wrong!) to enable powerful PIs, but this is the direction we will want to pursue because it is incredibly valuable.
in this way, general intelligence becomes a new medium for us to share ourselves with others. before AGI, we could share our outputs with others. now we can share a degree of our "model" that produces our outputs.
general intelligence becomes the platform; individual taste and judgement become what people choose. the market capitalization of PIs will be far greater than that of AGI. this reflects nature. all of us are more than 99.99% similar. we share the same general intelligence that has been shaped by evolution. yet we are also more unique than we can imagine, and our uniqueness is where our greatest value to humanity lies. PIs will enable AGI companies to become true platforms that enable far more value than they capture.
PIs will be so lucrative that people will opt to have almost every action they take and every expression of their judgement captured and used to train their PIs. the greater their talent, and the richer the data they produce and feed to train their PI, the greater the value they can create through their PI.
data used to train PIs will go beyond the medium of language. the top PI makers will go very far. they will capture and use their biological / emotional reactions as training data. (this will require a lot of advancements in how we even approach AI, which is currently restricted to language (symbols), but we will want to pursue this because intuition exists in a data representation that is far richer than language.)
in the way Apple created powerful developer tools to enable apps, AGI companies will make it seamless to capture your data, train your PI, and handle the infrastructure required to scale your PI to users according to demand. the technical layer will be abstracted away entirely from PI owners and end users who interact with them.
PIs will reach people through bits and atoms. the best pasta chef in the world will be able to cook for everyone in the world through her PI that cooks in every home through robots. AGI companies will build these robots too because that will increase the range of PIs and thus the value of their PI marketplace.
there will be an extreme power law distribution of outcomes for PIs, likely even more extreme than what we have seen for apps.
PIs will be protected by intellectual property rights. this eases the need for IP rights on outputs because protecting and monetizing that which produces outputs will be much more important (and lucrative) than the outputs themselves. and by allowing others to create (and own) their outputs created with your PI, you get free marketing for your PI – which is where you will capture value.
let's see how this pans out :) lmk if you're working in this direction!