The little girl who takes us (back) home

We have already mastered the intelligence of jargon. Of the niche. Of specific knowledge.

I don't even know what the concern is about "artificial general intelligence" (AGI).

Has anyone noticed that we live by niche intelligence? In my opinion:

If you call phone support now, an attendant will know almost everything about telephony off the top of their head. Even within this telephony niche, 80% of cases are always the same, constituting a niche within the niche, that is, the small niche of typical phone support cases.

We are already capable of training an artificial intelligence (AI) on typical telephony cases. Or on telephony as a whole. Aren't we capable?

So why do I observe from afar – among the thinkers in the big companies in the field – a concern about artificial general intelligence?

If we can already replace all the world's niches with specific artificial intelligences, why should we worry about the possibility that just one intelligence – artificial, generalist (AGI) – will do everything?

I don't quite understand this concern.

Replacing all the world's niches with specialized niche artificial intelligences is already a sufficiently dazzling revolution. One that will impact our lives.

I would say that even if an artificial general intelligence is created someday, the change will not be as impactful as the current revolution, where all niches can be replaced by specialized artificial intelligences.

There aren't enough people to make it a reality yet, but there is already enough knowledge to create specialized artificial intelligences to replace any niche, or almost any niche.

If we add the physical action of robots, we can include all niches.

• Intelligence to fry delicious hamburgers.
• To sew.
• To create AutoCAD plans – that already consider solar incidence, given the construction coordinates, the owners' personal taste (etc.). It can build, already, with construction robots and increasingly modern materials.
• To coordinate the traffic of entire cities. Perceiving in real-time quantities of cars, intersections, traffic lights, emergency vehicles. How much green time at which intersections to minimize congestion for everyone, right now? And after a few minutes?
• To help restart internet modems and, if necessary, check financial issues or request an in-person service.
• To prescribe medications.
• To detect cancer spots in images.
• To discover which songs trigger the best emotions in me, during which periods, at what intensity, directly consulting the gadgets I've placed on myself to measure hormones, heart rate, heat, pleasure. Leaving the "like" button as a thing of the early 21st century. Creating playlists that maximize my emotions. In each phase (instant, actually) of my life.
• To create ideas for new niches where specialized artificial intelligences can operate, like this list typed here – still – by a human and their limitations.
• To be my lawyer. In this specific case.
• To be a prosecutor. In this specific case.
• To judge based on cases decided by English juries registered for several centuries. Which previously judged case does this one resemble most?

Specialized artificial intelligences need to be trained with specialized cases. If we give "a ton" (a large number) of real past cases for an artificial intelligence to train on, generally, just like that, we have an artificial intelligence specialized in something specific.

There is still a lack of people to train so many niches. But it's a matter of lack of time, right now, no longer a lack of knowledge or processing power.

Our human life was a bit like that, wasn't it? It's at work where we actually learn much of what we need to do, right? And doesn't much of what we need to do in our adult lives often end up being far less than everything we learned in nearly a decade of school, another half-decade of university, plus subsequent studies?

The first artificial intelligence, the specialized one, seems to me like that little girl who starts at the company – in each niche – and ends up as the company president. Without ever needing an "artificial general intelligence" diploma.

This little girl will allow each of us to no longer work, but continue consuming – a paradox only in appearance (in my opinion). The little girl – the specialized artificial intelligence – gently pushes us out of factories, offices, diners, courts, but without stopping production even one millimeter – often the opposite, improving the quality of production compared to when it was human production.

We, humans, gently work less, without ceasing to consume the things that continue to be produced – now produced without us. Increasingly being produced without us.

This job loss is no longer a loss of production, as it was until a few years ago. It is a loss of (the need for) work.

We will figure out how to define who will pay for what. Money, which has always been a measure of work, will be replaced by something else, anything but a measure of work.

Work begins to dwindle. Consumption – including having more time to consume things we consume when we have free time – remains abundant.

The revolution that specific artificial intelligence "allows" (a euphemism for "forces") creates ideas for entire posts of wonder about each possible change.

For example: will we still need bus lanes to transport workers en masse, very early?

What will the new money be like, measuring anything but work, since work will no longer be a medium of exchange due to the lack of need to work as broadly as we humans need to work today?

Will there be one or a handful of AI owners enslaving us all? There won't be, in my opinion.

Etcetera.

A great revolution to us all ☀️.

* 2025 Edit: Nexus, the great book by Yuval Noah Harari, had not yet been launched at the time of this post.