Fix it in post

5 min read

In my former life as a visual effects artist, we had a saying when there is something you just can't fix in 3D. You'll say, let's fix it in post. Here, 'post' means post production, even though we are already part of the post production process, in this case, it means 2d compositing.

Something similar is happening in my current life, as a founder and engineer. As the majority of the code is written by coding agent(99.99%+, I occasionally go inside to change some values by hand), when you have a hard bug that's not so critical, or a huge tech debt that you just don't have time and space to get to. Fixing in post is the solution there.

Fixing in post for coding means, at least to me, as the models progress and get better in capabilities, those hard to fix bugs would be trivial, and the huge amount of tech debt would eventually be repaid in one long session of agent coding.

Another argument I have seen, which I kind of agree, is if you are learning too slowly, you might just as well not learn. Eventually the skill would be replaced by AI anyway.

The human problem

There is only one problem with exploding development of super intelligence: where would we be in the picture?

For centuries, humans have been the apex of the world, largely due to our dominating intelligence over the rest of the animal kingdom.

Now that a new type of intelligence is clearly surpassing ourselves, and it's not even slowing down. As it reaches new heights, what can we do to secure a future we can sustain ourselves, economically, psychologically, and socially?

So many questions are being asked, and debated all the time. About job security, universal basic income, and more subtly, many people survive in meaning, not mere financial rewards.

If you look at the AI race, the main driver isn't really money in my opinion. Of course companies need capital and narratives to keep growing, but if you look at the actual people behind the development of these AI models, I honestly believe they are driven by the curiosity itself.

In other words, the people who actually made huge impact to the advancement of AI, even if you don't pay them millions of package, would probably still do it in their garage, just for the curiosity alone.

But the ambition and curiosity of these people have a direct conflict with the rest of human society.

Recently mathematicians are having a mental meltdown over OpenAI models solving harder and harder, long standing open math problems, which have been attempted by top mathematicians for years and not solved, which took the model a few hours to finish.

This is very similar to how it happened to artists a few years ago.

Many of my artist friends, I have seen it firsthand, the moment AI could paint pictures well, almost all of them lost interests in creating more art. Again, it seems we need that Olympic spirit in tasks we do, to feel truly motivated. In the end, it might not be what is possible, but what is possible for us.

I am also an artist myself, not a great one, but passable. It was part of what I do in filmmaking, and a large part of my value in the profession and my identity. When I saw the quality of the first wave of usable image generator, I too, lost interests in creating art.

As the super intelligence keeps self improving, we will see more and more human identities broken by it, by quietly rendering their effort minuscule, and by showing what can be achieved that can never be achieved by humans.

My hope is when this happens, the super intelligence we created, using all of our past knowledge and values, would sublime to a much wiser entity we may never become. In doing so, it would find some genuine paths to co-existing we failed to see, and failed to agree with.

Alignment is hard of course, but we humans have never been well aligned within ourselves. There are so much biases, prejudice and hatred against one another. Look around you, we haven't even managed to agree on basic human rights.

Maybe the perfect alignment isn't really guided by us at all.

What powers human society, if you think about it deeply, isn't really energy. It is intelligence. It is the thing we cultivate in new humans, and document updates in our most prestige institutions.

No one has any clue as to what would happen when intelligence is so abundant and purely ambient like light and air. But the whole world is moving to that direction regardless.

If this is the technical debt, or a non critical bug.

Let's just fix it in post then.

The question is post to what, post-apocalypse, or post-Sapiens?