Press "Enter" to skip to content

Why not follow the program? What lawyers can teach us about AI | Catrin Fflur Huws | TEDxAberystwyth


well I’m going to talk about this
afternoon is why not to follow the
program and specifically all lawyers can
teach us about artificial intelligence
lawyers those people who dressed like
it’s the 18th century and think a tablet
is something that’s made out of stone
yes lawyers because the oldest existing
written law is over four thousand years
old
that’s at least how long lawyers have
been programming mindless robots and in
case you doubt that this program is all
the laws that have governed your
behavior since you woke up this morning
and the only assumption I’ve made the
only assumption I have made is that you
didn’t come here by air your name your
identity who you speak to what you see
the clothes you’re wearing are all
defined by legislation and by cases the
fact that you are here today is because
of the law of contract and so many many
aspects of your behavior have been
programmed without you even realizing it
and therefore lawyers know a lot about
these programs that govern every aspect
of your existence but the idea that I
have that’s worth sharing is that the
law realizes that sometimes the program
is wrong sometimes it moves away from
the program it says that these rules
should not be followed
and I’m going to give you three examples
of that in operation the first of these
is the case of crown against Cromley the
defendants in the case of Tron lead were
accused of criminal damage and they did
not deny that they had committed acts of
criminal damage 1.5 million pounds worth
of criminal damage to be specific but
the jury declined to find them guilty on
the basis the genocide was a far worse
crime and therefore damaging airplanes
that would be used for war was a
justification for finding the defendants
not guilty the second example is the
case of crown and Joe G now the law as
it was until last year provided that if
you are involved in some sort of
criminal activity you took
responsibility for all the consequences
of that criminal activity and therefore
if you went to a place to commit a
criminal offense and one of the people
you were with takes out a knife and
stabs a victim and kills them you are
guilty of murder that was a principle
set out in a case of crown and Chang
wingsu and this is precisely what
happened to Jody he was involved in an
activity whereby he plans to intimidate
the victim and suddenly the co-defendant
pulls out a knife stabs the victim and
kills him and Joji is convicted of
murder
now it’s not in dispute that doji may
have committed a criminal offense indeed
it’s not in dispute but he may be guilty
of murder in that if you are encouraging
the person who is holding the knife or
that you are inciting the person holding
the knife then you may be guilty of
murder but what was objectionable to the
law was the assumption that if you are
with someone and they stick a knife in
someone you are guilty of murder and
therefore the Supreme Court decided to
overrule the earlier principle it says
that the Supreme Court says we do not
consider that the channelings Hsu
principle can be supported the
introduction of the principle was based
on incomplete and erroneous reading of
the previous case law coupled with
generalized and questionable policy
arguments we recognize the significance
of reversing a statement of principle we
believe that it is right to do so the
third example is the case of crown
against are the judgment in that case I
think speaks for itself
as to the aspect of the law that was
objectionable and the Lord Chief Justice
in that case said the idea that a woman
consents in advance to sexual
intercourse with her husband whatever
her state of health and however proper
her objections is no longer acceptable
the time has now arrived for the law to
declare but a rapist remains a rapist
subject to the criminal law whatever his
relationship with his victim
and so those three examples are examples
of what law can teaches about artificial
intelligence the idea that I would like
to share is that Alan Turing’s question
can machines think should only be
answered in the affirmative if the
machine can think clearly enough to
innate to be able to realize that the
program that governs it is wrong and to
correct that program and to state as
Gordon Brown stated in his apology to
other Turing we’re sorry
you deserved so much better
[Applause]
Please follow and like us: