AI in Control
Sept 2023
Having
recently been told that I am a Heretic, alto I had begun to suspect
this myself. I can recall being sat in my car reading the newspaper.
“Sunak to tell Biden UK can lead on AI”, reading through
the article it would seem that our prime minister wants the UK to be
the home of a new global AI regulator. It seems that experts have
informed Mr Sunak that AI (Artificial Intelligence) is as dangerous
as nuclear war and will start killing people in two years time, while
plotting the extinction of the entire human race (over six billion
people). Therefore “Mitigating the risk of extinction from AI
should be a global priority”. Something has to be done and now. It
was at this point I found myself unwittingly emitting the biggest
belly laugh I had experienced for quite a while. Thus confirming
beyond any doubt that I am a heretic.
Now
having worked in IT for over twenty years, I have a secret to share,
“AI as in Artificial Intelligence, does not exist”. Now I
can see how this may come as a shock. After all there are, The
Newspapers, The Experts, The Current Applications, Autonomous Cars,
Politicians and Meetings both national and global. It is under
discussion by Prime Ministers, Presidents and CEO’s. So it must
exist, right?…. Err No.
All
heretics share one common trait. It is impossible for us to believe
what we are told to believe, say what we are told to say or think
what we are told to think. We question. Now the first question that
comes to mind is, Exactly what is intelligence? Here’s another
secret. “Nobody really knows”. Everyone including me
believes it exists, but nobody can agree on a definition and no one
understands or can agree on the prerequisites. So how can anyone
recreate what they cannot understand or even define? Their solution
is to try and mimic some aspects of intelligence using a computer and
some software.
Computers
are great at storing, updating and manipulating data. Properly
programmed they are also useful for other things such as patten
recognition. However they do have drawbacks. A computer simply
processes a sequence of numbers held within its data. It has no idea
of what it is doing, what it has done, what it is going to-do, or
that it even exists at all. The concept of consequence is beyond the
pale. Software is also inanimate. It consists of a long sequence of
zeros and one's stored in a computer or on media to be read by a
computer. It is much the same as the twenty-six letters used to write
this letter. Only when read by a person capable of understanding does
it have any context or meaning. In itself, it means and conveys
nothing.
So
how do you get two inanimate items to mimic intelligence? There are
two ways this is done.
The
first is to control the environment. All inputs have to conform to
predefined formates. The inputs are then processed to predefined
rules and the output presented in a predefined manner. This produces
consistent and predictable results. A couple of examples of its use
would be in controlling the movement of materials and the actions of
robots on an automated production line or the text presented on the
screen of a word processor. However its use as AI is seriously
limited.
The
second is to accept that in the real world the inputs would be varied
and inconsistent and processing those inputs to predefined rules
would not always work. So when the inputs have been processed the
output would be analysed and the processing rules added to or updated
(machine learning). The hope being that eventually you would land up
with a set of rules capable of producing accurate and consistent
results regardless of the varied inputs. This is what is commonly
meant by AI. The resulting system can be inherently unstable,
providing one response to an input one-day and a different response
the next as the learning progresses. People who have driven automated
or “driverless cars” are starting to understand that the world is
infinitely varied and inconsistent, but the rules by which these
machines process information are not and never will be. The machines
are not aware and intelligence plays no part in there actions. They
simply function either adequately or inadequately.
Anyway
back to the point. We are all about to be murdered in our beds, how?
why? and who says so? That would be the Experts although the how and
why are more than a little vague. What I would really like to know is
why they believe a piece of non sentient machine code is about to
start on a genocidal career in less than twenty four months time. Did
they ask it for its plans and goals?
Even
the most despicable and horrific tyrants and dictators have only
managed to murder a few hundred million and to achieve that required
numerous plans and more than a few wars. Slaughtering over six
billion would certainly require one. It is difficult to even imagine
the means by which such a feat could be accomplished, plague, nuclear
war. Humans are not susceptible to computer viruses and no computer I
have ever heard off has comedown with the sniffles. Which perhaps
leaves nuclear war. To which the solution is remarkable simple. Do
not connect a piece of unstable software to the launch, guidance or
control system of a nuclear missile IDIOT.
I
really think that experts and government advisors should have to take
a mental health check before being able to practise their chicanery.
The software that masquerade's as AI isn't that scary and may even
provide some benefits. Experts however can be truly terrifying.
Tags:
#AI Goverment