Grado de dificultad: 3 (Por el idioma y por el tema).

Columnista: Roberto

Hay una queja (justificada) en mi contra: que soy demasiado critico en mis análisis sobre ciertos temas.

Lo admito, y es la principal razón de la bajada de frecuencia de mis artículos en TMN. Muchos serían demasiado críticos, y no quiero que este blog se vuelva otra “Casandra”.

Sin embargo, hay opiniones críticas que valen la pena expresar, cuando los temas son relevantes. Escribir en otro idioma ayuda en suavizar el impacto local.

Artificial Intelligence is trending

Artificial Intelligence, interesting subject

Artificial Intelligence is an interesting subject, but it bothers me (for the lies that are told about it).

It often appeared in TMN’s columns, though. Here is a list of our most recent ones:

It is a sign of how much it interests us.

But there is no way any of us would risk our lives letting a Tesla vehicle drive us “autonomously”.

A wrecked Tesla - Origen desconocido

Intelligence, the missing notion

Preliminary disclaimer: I am nobody (therefore not intelligent). That being said, I have (a lot of) questions on the subject of A.I.

Note: Microsoft’s grammatical correcting A.I. tells me that “that being said” is an incorrect a non-optimal way to write. Sorry, A.I.!

I recently read a remarkable piece from The New Yorker:

Can Computers Learn Common sense?

It is remarkable, but not for a good reason. One major notion is missing in it: … Intelligence.

I was amazed by its contents, starting with the notion of “Common sense”, which existence I always doubted.

The Common sense trope

A few quick questions

In the article, they say:

The A.I. answered the questions in a common sense way seventy-two per cent of the time, compared with eighty-six per cent for humans

Question-answer-instances from five commonsense benchmark

A quick question: why 86 % (in average, I presume) and not 100 %?

Which suggest others:

  • What questions are they using to assess this percentage?
  • What result would we, individually, reach (assuming that we are considered human, or course)?
  • Then, most importantly, who decides what “Common sense” is?

Do you remember how we all were selected (rather than educated) at school? And how we were all put in labeled boxes?

Well, the same people that were labeled as superiors are the ones that created this genius concept … Without experiencing any unconscious bias, of course.

What is “Common sense”?

I don’t think there is such thing as common sense. It is a trope, just like “responsible adults”.

It is a consequence of social conventions, which differ from Los Angeles to Tokyo. Some differences are subtle, others will be bigger.

Further, each of us has his/her own understanding of it.

Then, on top of it, comes an additional interpretation by the superior beings that create the criteria for the I.A.

Now, we know why the “recommendations” of Netflix, YouTube, or the adds that collect our data, are so sloppy.

What about reasoning, instead? It would be more learning-friendly.

The illusion of intelligence

What the article really shows is how crude is the understanding of intelligence by the people who pretend to emulate it.

But this is not what it is about. It is the good old Silicon Valley tactics:

Fake it until you make it

There is, however, a question in this case: how long is the faking going to last?

There is a famous case: a supersonic commercial plane, the Concord, started to fly in the seventies. it was then predicted that it would soon have challengers, with better solutions.

The prediction was periodically repeated, since, but still no successor in sight, 50 years later.

So, yes, the tale about A.I could last a long while. And that is unfortunate because it looks like brilliant minds are stuck in an illusion: that of their intelligence.

Artificial Intelligence and common sense

Header Kirabytes1