 |
| Ramón López de Mántaras |
In an article published on 3/22/2021, these words are attributed to Josu Bilbao, head of the ICT area of IKERLAN:
In
three years, artificial intelligence will make an intelligent dialogue with
machines possible.
We have seen
many similar predictions since the term Artificial Intelligence (AI) was invented over 60 years ago. In most cases (if not all) those
predictions have been too optimistic. Is the same going to happen here?
I have consulted
with one of the world's leading experts in the field of the automatic analysis of
natural language, a technique used to implement the type of applications
referred to in the article I am commenting, and he told me this:
It depends on how "intelligent dialogue" is defined. If it is restricted to a specific domain, for example, controlling your home automation systems using Google Home, that can be done now with a success rate above 70%, while ten years ago it was science fiction. If you want to have a philosophical conversation where your interlocutor understands what he is saying, rather than generating texts from a language model, this is too optimistic.
Almost at the
same time (just one week later, on 03/29/2021) one of the most important Spanish
names in the field of Artificial Intelligence, Ramón López de Mántaras,
director during many years of the Research Institute on Artificial Intelligence
of the Spanish National Research Council (CSIC), with many awards of the
specialty to his credit, is much less optimistic in an
article published in the major Spanish newspaper La Vanguardia about the
possibility of achieving artificial intelligence comparable to human.
These supposed
advances in AI, always about to be achieved, according to some self-appointed
experts such as Ray Kurzweil, who has been insisting on this since the 1980s,
would bring us in about a quarter of a century to what has been called the technological singularity. The result: machines more intelligent than man in
all fields of activity, which will either supplant us, or merge with us,
forming hybrid creatures, half biological, half technological, which may achieve
some of the goals most desired by humanity since the beginning of time, such as
immortality. This is what is called transhumanism.
Faced with
these ideas, López de Mántaras says that it’s true that the technological
development of hardware has taken place with an exponential growth (Moore's
law), which until now has been quite approximate, although it shows clear signs
of having entered into the decreasing phase of the logistic curve. But such
growth has never taken place in the field of software, where it should have happened
too, if the exaggerated predictions being made should be feasible.
These are some of
the things that López de Mántaras says in his article:
The
main hypothesis is that there is exponential progress in the field of AI,
which, in my opinion, is highly debatable... The algorithms currently being used
in what is known as deep learning (the most successful and important current
trend in AI) are over thirty years old and, although they have been improved in
some aspects, conceptually we can say that they have not progressed
significantly since then. It is undeniable that, in recent years, there have
been important results in AI, but this has not been due to a great progress in
AI algorithms. The reason has been the availability of large amounts of data
and high-performance hardware to train them. On the other hand, these results
have been exaggeratedly amplified by the media - and also by some of their
designers - which has led to the creation of unrealistic expectations about the
current state of AI.
When talking
about AI, in the press, and sometimes also in scientific literature, some terms
are used incorrectly, spreading confusion among non-specialists. Let's look at two
of those terms:
•
Consciousness: The Spanish
official dictionary, in its first acceptation, defines this word thus: capacity of a human being to recognize the
surrounding reality and to relate to it. If we intend to attribute consciousness to intelligent programs, this
definition should be redefined as follows: capacity
of a being to recognize the surrounding reality and to relate to it. In this way it would apply not only to human
beings, but also to other living beings, and potentially even to machines. It
is evident that many living beings and some robots are capable of getting information
and relating to the surrounding reality. In this sense, it would be feasible to
apply to them the word consciousness. But only in that, very restricted, sense.
•
Self-consciousness: The second
acceptation of the word consciousness in the Spanish official dictionary is this: immediate or spontaneous knowledge that a
subject has about itself, its own acts and reflections. In other words: self-consciousness is the idea of
the self, the consciousness of being a person, closely related to free will and
responsibility. In this sense, humans are the only beings about which we know
with certainty that they possess self-consciousness. Any application of this
term to programs or machines, however much one may speak of artificial intelligence, is an abuse of language.
As López de
Mántaras says in his article:
Algorithms
understand nothing... Humans, unlike artificial intelligence systems,
understand the consequences of our actions and decisions... in my opinion no
matter how sophisticated AI becomes, it will always be different from humans.
This is what I
have been saying in this blog for years. I am glad of this agreement between López
de Mántaras and myself. In general, this belief is shared by most experts in
artificial intelligence. Those who make outrageous predictions are usually those
who don't know much about the subject.
Thematic Thread on Natural and Artificial Intelligence: Previous Next
Manuel Alfonseca
No comments:
Post a Comment