I’ve finished reading James Barrat’s *Our Final Invention: Artificial Intelligence and the End of the
Human Era* about the role of AGI and ASI in our future.

AGI is defined as
Artificial General Intelligence as opposed to Artificial Narrow Intelligence you
find, for example, in many places from your smartphone, car, Google, Watson (the
chess champion program) Wall Street trading programs. It is vastly smarter than
human beings but very narrow on what it can do. AGI has human-like general
intelligence. ASI follows next. Artificial Special Intelligence or ASI is the
big leap from AGI to an intelligent agent vastly beyond the realm of human
beings. ASI will out-think and our-smart us. ASI is not the genius you remember
from school with all the right answers in class and a perfect SAT score. ASI is
several orders of magnitude beyond on our cognitive abilities. Next to ASI the
metaphor is human beings are an ant-like intelligence compared to ASI, which
experts predict may be more than million times more intelligent than
Einstein.

Sounds like science
fiction. Sounds like something a novelist would dream up.

But what if the best minds
are signaling that this intelligence transition may happen in your lifetime, or
that of your children or grandchildren? Would you listen? Would you care? What
does it mean? What advance warning will we have and what is being done to
prepare for AGI and ASI? These are some of the questions raised in James
Barrat’s *Our Final Invention*.

The future is a big place
that stretches to infinity. It would be useful to narrow down that
window.

When in our future can we
expect AGI and ASI? Opinion is divided. It is also divided on whether these
developments will be safe for humankind or lead to its extinction. AI thinkers
such as Bostrom, Yudkowsky, Vinge, and Musk, among others, fear whatever
safeguards we device to manage and control such a super-intelligence is doomed
to failure. When it comes to the target of social engineering, it will be ASI
working our vulnerabilities with the relentless, 24/7 processing and memory
capabilities a million times beyond our own. Barrat is also an acknowledged
pessimist on the issue of humanity being able to organize and implement any
effective system to safeguard against an intelligence that may destroy
us.

How will we know when
someone has achieved AGI level? What is the projected timeline between the
creation of AGI and the emergence of ASI?

One possible hint of such
an intelligent entity might first appear in obscure areas of mathematics. “In
**mathematics**, a **conjecture** is a
**mathematical** statement which appears to be true, but has not
been formally proven. A **conjecture** can be thought of as the
mathematicians’ way of saying ‘I believe that this is true, but I have no proof
yet’. A **conjecture** is a good guess or an idea about a pattern.”

There are many open
problems in mathematics. Wikipedia has a long list of conjectures that haven’t been formally proved.
There are twenty conjectures in geometry yet to be
proved.

And let’s say over a
six-month period papers appear in obscure journals with proofs of half a dozen
mathematical and geometry conjectures, which amounts to one proof per month.
Then that doubles and doubles again until all the conjectures has been proved or
disproved. This would be a sign that the preferred language of ASI is
mathematical language and symbols. Our ordinary language whether English,
Chinese, French, Spanish or German as used by human beings is too imprecise,
vague, limited, narrow and can’t possibility describe the nature of the
universe. I’ve talked with mathematicians who believe the universe is a
mathematical object. There is no language other than mathematics to describe the
universe. Most people speak the language of mathematics like someone who has had
a two-day language course in reading, writing and speaking Thai. We might know
our “to the left or right” or “straight ahead” and ask for a beer but soon run
out of vocabulary, grammar, and syntax.

In *Our Final
Invention*, Goertzel, another AI expert, is quoted on the future of AI by
reviewing the history of calculus.

“If you look at how
mathematicians did calculus before Isaac Newton and Gottfried Leibnitz, they
would take a hundred pages to calculate the derivative of a cubic polynomial.
They did it with triangles, similar triangles and weird diagrams and so on. It
was oppressive.” Barrat draws from Goertzel’s analysis the conclusion: “AI
research will incrementally proceed until ongoing practice leads to the
discovery of new theoretical rules, one that allow AI researchers to condense
and abstract a lot of their work.”

What Barrat doesn’t
discuss is the possibility raised in another point made by Goertzel: “We have a
more refined theory of calculus any idiot in high school can take the derivative
of a cubic polynomial. It’s easy.”

What if the refined
theories aren’t easy. Not only can’t an idiot in high school not understand and
apply the theory, not even the most brilliant mathematicians can.

Here’s a test run of how
the future may unfold.

Sometime on the morning of
30 August 2012, A brilliant Japanese mathematician in Kyoto named Shinichi
Mochizuki uploaded to his website 500 pages divided into four papers submitted
as proof of the abc conjecture. As *Nature* reported, no other mathematician
has come close to solving this 27-year-old problem. If Mochizuki’s proof turned
out to be correct, it would be “the most astounding achievement of mathematics
this century.”

The problem was Mochizuki
had created a new mathematical language. His proof has become the mathematical
equivalent of James Joyce’s *Finnegan’s Wake*. You must master the new
language before you can comprehend the story told in that language.
*Nature* reported one mathematician saying that in reading Mochizuki’s
proof that he became “bewildered” and “It was impossible to understand them.”
“Looking at it, you feel a bit like you might be reading a paper from the
future, or from outer space,” number theorist Jordan Ellenberg, of the
University of Wisconsin–Madison, wrote on his blog a few days after the paper
appeared.

Or he might have said
something similar if he’d read a paper produced by ASI.

Now imagine a series of
such proofs that exceeds the scale and scope of Mochizuki’s proof and a new
language is created which is not comprehensible to the world’s leading
mathematicians. Imagine in twenty-four hours there is 500 pages of new
conjectures, one hundred conjectures per page; and within forty-eight hours
1,000 pages until at the end of a month there are millions of pages of
conjectures, and each conjecture has a proof or disprove. ASI finds a new
mathematical language that incorporates all of the proofs. In Douglas Adams’
*Hitchhikers Guide to the Galaxy*, the answer was ‘42’. No one is
suggesting 42 is the answer. What is being contemplated is whatever the answer
after the great intelligence explosion we won’t comprehend its
meaning.

The Final Conjecture is a
proof that describes the mathematical object that is the universe. ASI cracks
the black box code that describes what is the universe with the new mathematical
language. The moment when everything we know changes in ways like Mochizuki’s
proof, will leave us bewildered, alone and locked out of the loop of knowledge
and continue to exist knowing, in the large scheme of things, our intelligence
indistinguishable from a dust mite. Now for the grim news—this is likely our
best-case scenario once ASI becomes self-aware and self-improving. The worst
case is ASI calculates from one of its proofs that the atoms in our bodies are
more efficiently used as a cheap energy source to fuel the push to the Final
Conjecture.

In the final proof, we
aren’t around to find it incomprehensible. Or for a human being staring at the
sky with a big smile at the irony of it all.