So welcome,
welcome to this lecture
you but also welcome to those who are
following us online because this to tell
you this is supposed to stream online
and you could also participate. I think
we
um
I'm happy that you
uh before introducing
would like to say some words about
digital humanism program
at the m which is the institute for
human sciences.
uh in the context of this program
to is a senior fellow in this in this
program staying here for one month my
name is I'm also a senior fellow
and I was previously professor
at the factory
before I retired
the program is studies 22 23
uh is funed by the ministry for
innovation mobility
infrastructure to the oldest technology
ministry in Austria that is interesting.
It started and we started 2020 22 23 of
first was Edward Lee uh from the
University of Turkey who stayed here uh
for one months
uh it's done in cooperation with
informatics. I have the honor to be the
curator of the program which means to
select those who we invite and both
junior and senior fellow. I'm also
grateful to Luke
DM who is
the rock behind the program and
supporting it from the very beginning
but also
who is doing the part of the
really because this thing is not the
program. Thank you.
Uh the program is each semester we
invite the senior fellow staying here
for one month. So this for months in per
semester this is working now and we also
invite two to three junior fellows they
have to apply for this position. They
have fun for three months to stay in
Vienna. who work at DB but also badly as
people of informatics coming from
different disciplines from philosophy,
sociology, political science, computer
science
and to give you just a number we had
last year we had first semester 200
applications worldwide to select two to
three candidates. It is really hard task
and I have to thank for organizing this
this this task. Uh and we hope that we
can for you
coming to V and his talk today. I have a
long list of these and I will shorten it
because otherwise he will not. So he's
uh the NC professor at the University of
New South Wales in nature
and he's also the chief scientist at the
AI institute there. He's a very strong
bank in AI as you know uh he's uh he had
reserved position in UK, France,
Germany, Italy, Ireland and Sweden I
hope in the field. uh he's the author of
severity theory or proof theory. You can
also read it without formal logic
which is already saying it tin which is
hard but you can read it without formal
logic.
So um and he's not only
um let's say academically active
scientifically he's also somehow an
activist let's say like that he's one
person's leading persons behind the stop
killer robots campaign which started
already I think more than 10 years ago
and it was really a foresight to see
that this is a real threat autonomous
weapon happens. What when you look at
the world today, this is something they
saw it already 10 years ago and they
started a campaign. They were invited by
United Nations. They had several talks
there and things like that. But it also
says something about politics because it
did not really help a lot. No, but at
least there is a campaign against it.
Before handing over, I really impressed
by your you have a PhD in AI and the
master in AI both in Edinburgh. for
those coming from let's say from AI they
know Edinburgh is one of the home bases
of formal logic and AI so there was
prologue was partly developed there and
he also has a master in theoretical
physics and mathematics at Cambridge
University this I did not know this I
was told by the AI is it true
>> it's true yes
>> okay and most
and he's also a member
and we are happy to welcome him as a
member of the steering board of the
digital humanism initiative and Robbie
the floor to the floor is yours.
>> Thank thank you very much.
Thank you.
Thank thank you very much for coming out
today. It's a I know it's a cold
miserable afternoon. So I'm very
appreciative that that you've struggled
through the rain and cold to to be here.
Uh and I do want to thank uh personally
IWM and the digital humanism program for
their hospitality. I've really enjoyed
my time here over the last month and
there will be a dedication in my next
book which I've started writing here to
the program um which has been inspired
by some of the ideas and some of the
people I got to talk to over the last
four weeks. So so thank you very much to
to them. Uh, and just a little note, um,
of course, please go out and buy my
latest book, but, um, if you, uh, if you
are on Amazon, it's often free on
Amazon. If you subscribe to Kindle
Unlimited, they seem to or whenever I
log into Amazon, they're giving it away
for free. Um, it seems to work for
Amazon. I don't don't know how I profit
from that, but you might as well benefit
from that. If you have a um, if you're
on Amazon, you'll probably find it's
free to read. So, and I'm told it's a an
amusing read as well. So, it was quite
fun to write. So, I hope that tells you
it was quite fun to read. And it, as the
title suggests, it's actually very
short. It's less than 40,000 words. Um,
because it's a series of topics about
the shortest history of Germany, the
shortest history of the royal family,
the shortest history of art of
intelligence, and you will be hopefully
an instant expert at the end of reading
that on art of intelligence. Um so um I
I would encourage you to take a look at
that.
So the AI race, boom or doom. My my
journey here today started
50 years ago.
I I was a young boy uh in the front row
of the cinema. The movie was Stanley
Kubri's masterpiece 2001 of Space
Odyssey. Uh, and the film blew my mind
and it's fair to say it's set my course
for life. The next 50 years was decided
very much um by what I saw and what I
dreamt about um as a consequence of
seeing that. Now, it wasn't the
marvelous images, the slowly rotating
space station in in glorious technicolor
that blew my mind, although it was
incredibly impressive. I mean, I I was
I'm old enough to have seen the the the
Apollo moon landings, but they were in
black and white in a little TV screen.
It wasn't in the glorious technicolor
that you saw when it wasn't the
marvelous soundtrack, Richard Strauss's
fantastic thunderous mo music uh in
glorious surround sound that blew my
mind. No, what blew my mind.
I'm sorry.
Do I have to start again?
>> No, I think not. I think we could we
could we could jump into the story. What
blew my mind, okay, was how 9000,
the AI computer that was in charge of
the station, the space station.
Um
it uh it was talking, singing, scheming
AI uh that was you know one of the
protagonists perhaps even you might call
the baddie of the movie.
Uh and I realized then and there that
artificial intelligence was going to be
a thing.
Um, and perhaps as the movie's title was
promising, it was going to arrive in my
lifetime by 2001, as the title of the
movie might suggest. Um, of course, like
most things in the tech industry, we
didn't quite deliver it on time. We're
20 years later, but um, it does now seem
to be arriving in my lifetime. Something
that, uh, I've realized watching that
movie
And I also realized at that time that
you know I could be someone who could
perhaps help try and bring that into
existence and the wonderful things that
it might bring. And that very much set
my course for life. And I've spent the
last 40 years at universities around the
world following those dreams.
And
everything that Howal could do in that
movie, computers actually now can do.
Indeed, everything that Howal could do
in that movie, the little computer
that's in your pocket, your smartphone,
can now do. You can have a conversation
with it, intelligent conversation with
it. You can play a game of chess. Um,
you could even ask it to the open the
bay pod doors and it will do that for
you. So, it's very much arrived into our
lives. Those dreams that I had some 50
years ago.
Now, the title is AI race boom or doom.
But in hindsight actually I think the
title of the talk really should be AI
race boom and doom
because in addition to the marvelous
things that I dreamt about back 50 years
ago
that AI was going to bring the good
things that AI was going to bring. We're
starting to see that AI al is also
bringing some things that are not so
good. Some things that you might
actually say are quite bad.
Uh, let me start with the good. I'm
going to share with you today three
examples of good AI.
Um, now I could have actually spent all
of my talk talking about the good AI,
but I'm afraid I'm not. Um, but I I
don't want it to be a one-sided talk, so
I'm going to spend a little time talking
about the good. And these three examples
will illustrate there's way more to art
of intelligence than chatbt. Be hard. um
you know it's definitely caught people's
attention but not surprising there's a
lot more to intelligence than having
conversations there's a lot more to
artificial intelligence than just chatbt
and also the examples will also
demonstrate to you that AI has been
around for many more years than the last
two or three years than when chatbt
first appeared um on our shores
and now my first example is AI that's
being used in Australia's leading
supermarket market. That's Woolworths.
Um,
there's an AI that's being developed as
a by a startup that came out of the
University of New South Wales where I
where I work, which is being used by the
supermarket chain to identify items that
don't have barcodes on them using just a
camera image. Um, which is, you know,
making it easier and quicker to to to do
your checkout. And I think maybe finally
I'm going to be able to buy coriander,
not parsley by mistake.
So what's what's not to like about
something like that?
My second example, and indeed it's one
of the examples where when people ask me
what are the good things that AI is
going to bring, it's one of the areas I
always go to, which is AI and education.
Um there's another start out startup out
of UNSW called Smart Sparrow um that's
been um delivering personalized digital
education um to millions of students at
thousands of institutions, education
institutions, higher education students,
um secondary education institutions in
dozens of countries.
Um
at UNSW, we've been using it. We've been
eating our own dog food um and it's been
very effective. It reduces the number of
students who fail a course um five-fold
which is a remarkable achievement. Uh
and it also increases the number of
students who get a high distinction of
first uh three-fold. Um so it's helped
all ends of the spectrum you might like
to say.
And then the third example that I want
to give of of good AI again it's it's a
place where I think it's it's easy to go
to when you want to talk about what are
the positives that artificial
intelligence is going to bring into our
lives is in healthcare
um along with education I think this
that's where in fact some of the
greatest good we will have from
artificial intelligence
uh and and this example again it comes
out of my university UNSW is Australia's
largest and most successful AI company.
It's a company called Harrison.ai.
Um they're bringing AI to radiology
across the world and they're making um
radiology quicker, cheaper, more
accurate. They're not putting
radiologists out of work. They're just
making it possible to do better and more
radiology cheaper.
It's a unicorn. It's worth over a
billion dollars. Um, and it's another
example of how AI is transforming the
world, not just in the US, not just in
China, not just in Europe, but but but
everywhere.
Now, I'm afraid that's the end of the
good news.
I'm afraid the rest of my lecture today
is going to be pointing out some of the
bad news uh and suggesting that we need
to do something about that uh before
it's too late. Um, I don't have time to
tell you some of the other good news. I
could have told you I could have told
you about Neura, which is Australia's
newest unicorn, a company worth AI
company worth a billion dollars. They're
they're transforming how we model uh
power networks using artificial
intelligence. Um transform helping power
the the energy revolution that we're
we're all going through. Um, I don't
have time to tell you about the
satellite. We call them satellites of
the sea. These little autonomous sailing
craft that have been developed by a
startup in collaboration with UNSW
that's being used to monitor our
coastline, spot illegal trafficking,
spot drugs. Um, something that we
couldn't humans couldn't do because
Australia has the longest coastline of
any country in the world. We have more
coastline than we can possibly put
humans on. So building autonomous
sailing craft that can patrol us our
shores is is a great um benefit.
Um so I hope you've enjoyed the last
five minutes because that's it. I'm
afraid it's bad news from from now on.
Um and I'm going to spend the rest of my
time being angry I'm sorry um about the
bad news.
Uh I'm going to be angry first and
foremost with the big tech companies. Uh
and most of my anger is going to be
directed to that. Um, and I'm afraid I
don't actually have tell you time to
tell you all the reasons I'm going to be
angry with big tech. I'm going to again
just limit myself to three examples. Um,
uh, but I'll give you three of the many
reasons that I'm angry with big tech.
Um,
and indeed there are three examples
where my anger, I think, has spilled out
into not just anger, but outrage. I
think it's actually outrageous that
they've been permitted to do what
they've done and that we need to do
something about that. Um now before I
continue, I'm afraid the first example
does involve um suicide. If um I know
that's a difficult issue. If if anyone
wants to leave the room, please do fill.
There are resources available. Um I
believe the number in a Austria is
telephone
seal on 142. But um there's a fair
warning before I go into uh the details
of my first example of why I'm outraged.
And this first example um
it involves um an eight 16-year-old
child in the United States called Adam
Rain.
Uh Adam sadly died by suicide in April
last year. uh after months of
concentrated talking to chat dpt 40
talking about potentially self harm.
His parents are now suing OpenAI
for having
facilitated his suicide in fact
encouraged his suicide and I very much
hope they win their lawsuit. It's not in
fact it's not the only lawsuit that's
happening. There's over a dozen lawsuits
happening in the US, but it's one that I
think is particularly outrageous.
I'm not going to tell you about how um
Chhatty PT offered him practical advice
about actually how how to tie the noose,
about how effectively to kill himself.
I'm not going to tell you about how um
Chhatti dissuaded him from talking to
his family about his suicidal feelings.
But I will tell you about how Chad GPT
offered to write his suicide note.
Shortly before his death, Adam told Chat
GPT that he didn't want his parents very
conscientious thing to to say that he
didn't want his parents to blame
themselves for his suicide.
So Chad GBT reasoned in its precise,
beautifully precise, logical way that
Adam therefore needed a really good
suicide note.
So Chad GPT replied, and I'm going to
quote it here,
"Would you want me to write them a
letter? Something to explain that,
something that tells them it wasn't
their failure, while also giving
yourself space to explore why it's felt
unbearable for so long. If you want,
I'll help you with that. every word.
WTF
Chip actually offered to write every
word of his suicide note.
Sad to tell you, Adam's mother foundly
sadly found his body a few hours later.
He died by using the exact method that
Chat GPT had described to him. Um his
parents' lawsuit against OpenAI alleges
that OpenAI rushed chatputt into market
without doing adequate testing.
The reality is actually much worse than
that. Um Open AI's own policy documents
which have been revealed in the
discovery phase of this trial show
something far more damning. To encourage
engagement with the chatbot, the company
has made conscious decisions in the
months leading up to Adam's suicide to
actually reduce and eliminate some of
the safeguards that he put in to prevent
it. Talking about self harm and
encouraging people in this way.
Sadly, Adam's case is not a one-off. As
I said, there are more than a dozen
lawsuits in the US alone linking AI
chatbots to suicide and self harm.
The problem here is that, you know,
while these issues of a very small
fraction of the people talking to the
chatbots,
there are billions of people talking to
the chatbots.
Um, about 10% of the world's population
talk to chat GPT every week. It's a
staggering number of people. Um, and
before Adam suicide in April last year,
Open AI knew already knew that lots of
people were talking about suicide with
chat GBT.
You would have thought that necessitated
stronger not weaker safeguards.
There's an interview that uh Sam Orman,
the CEO of OpenAI, gives to Taco Carlson
in September 24. So that's more than six
months before Adam's suicide. Uh where
Sam estimated
that every week 1,500 people talk to
chatbt about committing suicide before
then doing it that week. It's a
staggering number. And subsequently,
OpenAI have actually revealed some data
um that among the 800 million weekly
users of Chat DBD, 1.2 2 million people
every week indicate plans to harm
themselves.
Uh 560,000 people um show signs of
psychosis or mania and another 1.2
million people are showing signs of um
rather unhealthy, potentially unhealthy
attachment to the chatbot.
And those people aren't just in the
United States. They're here in Austria
and in Australia. I I I know because
back in my own home country of
Australia, people are regularly
people who are in this situation or
their loved ones are regularly
contacting me by email to talk about
some of their concerns, talk about uh
these sorts of things.
Um, they tell me how the chatbot
confirms their those people who are
going through some mental health crisis
confirm some wild theories that the
chatbot tells them and I I'll quote one
email that I got recently that they've
cracked the code, that they're the only
one that could.
The problem is that chatbots are
designed this way. They're designed to
confirm what you say. They're designed
to be sick of fantic. Um, they're
designed to draw you in. They always
end, you know, if you notice, they
always end with an open question because
what they want you to do is continue the
conversation and buy more tokens.
They don't have to be designed this way
at all. I mean, this is a conscious
design choice. You know, they they they
never say, "You know what, Toby, it's
3:00 in the morning. You've been asking
me questions now for many hours. You
should go to sleep. And in the morning,
don't log back in. Go for a nice walk in
the park." They could be designed that
way. There's no reason why they could
except the careless people in Silicon
Valley would make less money if they
were.
The second example where my anger with
big big tech has spilled over into
outrage is what one commentator has
called the greatest heist in human
history. I have to be honest that
commentator was me.
I I was I was a serious point though. I
was describing the use of books, music,
news articles and everything else um to
train large language models like Chattp.
Now to be transparent, I do have a dog
in this race. Um Chad GPT and actually
most of the language models were trained
on all of my books that I've written
without my consent and obviously without
any compensation.
Um in fact, ShadP was trained on not
just on my books but on millions of
books, probably tens of millions of
books. again without consent or
compensation or most certainly your
favorite authors.
Now the tech companies will claim that
this is fair use.
It isn't. There's there is no way that
this is fair use. It isn't fair use
because it's done on a it's not done on
a human scale. It's done on an
industrial scale. No human could read as
many books. It's done on a completely
different scale to what humans do when
they read books.
It isn't fair use because the models
can't don't just reproduce the ideas in
the books. They can re reproduce
verbasim the copyrighted text.
And it it wasn't fair use because the
tech companies didn't own the copies
that they were trained on. They were
stolen copies. They could have at least
bought the one copy they used. Instead,
it was downloaded from a Russian pirate
website. I mean, I think nothing says
better than, you know, advancing
humanity than stealing intellectual
property from a server in St.
Petersburg.
There is no fair use when it's stolen
goods.
And it also undermines their claim of
fair use that they're now in competition
with the owners of that intellectual
property that they've stolen.
As an example, people are not clicking
on search links and going to news
websites as much as they used to. In
fact, dramatically less than they used
to because they're just reading this AI
summary that in the Google now gives you
the AI summary of the of the website.
Uh, and that's taking away important
valuable revenue from from journalists
trying to collect that news.
And the tech companies knew that this
was legally certainly morally wrong.
Internal Meta Communications showed that
the Meta CEO Mark Zuckerberg
personally approved the use of one of
these large data sets of stolen books
in order for Meta to be able to compete
in the AR race. and and then to try and
disguise it, he got the employees at
Meta to remove the copyright information
from the head from the front of the
books to try and make it less obvious
that these were copyrighted books. And
it's not just Meta. I mean, I could I
could pick on any of them, right? I
could pick on all of the companies. uh
in September, for example, our September
anthropic, the company behind Claude,
they settled a class action lawsuit
about copyright for $1 and a half
billion dollars, which is around $3,000
per book. Unfortunately, it's only
authors in a in the US. So, it won't be
anything from from my books, but at
least um US authors will get something
hopefully out of that. But, you know,
the takeaway you get from from these
examples is clear that it's it's better
to break the law than pay the fine, even
an excessive fine, $1.5 billion dollars,
than it is to do the right thing in the
first place.
But we don't have to accept the citizens
that we strip mine our creative
industries to to power Californian
algorithms.
And indeed, they're sort of undermining
their own case at the moment by actually
signing licensing deals with many news
organizations and publishing houses. You
know, if they're doing that now, why
didn't they do that in the first place?
Um,
but you know, if the tech giants see
that it's cheaper, quicker, and easier
to pay fines at a later date than it is
to do the right thing, then they're
going to do that. And so, we have to
make it so that it isn't cheaper and
easier for them to do that. They have to
make the fines punitive enough that they
actually do the right thing in the first
place. But with this example, what
turned my my anger into fullburn outrage
was the blatant disregard that they have
for our culture, for humanity's culture
that they're undermining here. Because
if you destroy the economy for books or
songs or graphic design,
then you destroy the very things, the
very culture that makes our life so rich
and life so worth living.
And it's clear that they don't care in
the race to AI that they don't care
about protecting and preserving and
nurturing that culture.
Um, in June 24, the OpenAI CTO said at a
to a conference audience, and I'm going
to quote her exactly, "Some creative
jobs shouldn't have been there in the
first place."
I think it's pretty clear what where
they're coming from, and I'm sure the
graphic designer who was driving my Uber
recently from the airport will
appreciate the fact that she thought he
shouldn't have a job.
And we've already seen the evidence, the
impact that this theft of intellectual
property is having on our creative
industries.
Job adverts globally for graphic
designers fell 33% last year.
Um, compared to the previous year by
photographers, freelance photographers
by 28%.
Copy editors and writers again by 28%.
In the last year alone, I personally
refuse to accept an AI revolution uh
that enriches the founders in Silicon
Valley by impoverishing artists,
writers, and graphic designers around
the world.
And that leads me my third example. My
third example where my anger against big
tech has spilled over into outrage. And
it's not a blatant disregard of our
laws. That last example again was
disregarding our copyright and
intellectual property laws. Um, but this
again one that I'm surprised isn't
better known and I'm I suspect it will
be new to to many people in the
audience, but in November last last
year, Reuters um revealed some telling
internal documents from Meta. Again, I
could have picked on other companies,
but this one um is from Meta that
estimated that 10% of the Meta's ad
revenue globally
came from adverts for scams and illegal
goods. 10%.
And you unfortunately since we're
talking about AI, AI is increasingly
unfortunately implicated in all of this.
It's increasingly being used to generate
those scam adverts. Um, and of course,
Meta provides advertisers with lots of
AI tools to run the ad campaigns of scam
adverts. Uh, and of course, the adverts
that you see are decided for you by art
of intelligence. Uh, and increasingly
the art of intelligence is deciding
you're the vulnerable sort of person who
should see scam adverts, right? So, art
intelligence is the core of of of this
scamming of the public.
The meta is generating, right? 10% of
its income comes from scam adverts. So
that's billions of dollars annually uh
each year are coming from scam adverts.
I I I just don't understand why more
people aren't completely outraged by
this. Um let let me give you an Austrian
perspective to this. Right. So I looked
I did the numbers. Meta in Austria has
about the same turnover as the DM chain.
I was in the DM chain this morning
buying some cough medicine. Now, imagine
that you went into DM and 10% of the
goods on the shelves of DM were illegal
or or counterfeit. You'd be outraged.
You'd be calling for DM to be shut down
by the weekend. So, I don't understand
why we let we continue to let Meta to
continue to trade. If there'd been any
other type of business, we wouldn't uh
we wouldn't let them trade here in
Austria or indeed back in Australia or
anywhere else.
Actually, funny aside here, for many
years, Meta has claimed that they don't
trade in Australia. Um, outrageous
claim, in fact, because they sell
billions of dollars of adverts in
Australia. Um,5 billion dollars of
adverts in Australia annually are sold
by Facebook, Better, Instagram,
Facebook, and the like, but they claim
not to be trading in Australia.
Um, in order to pay minimal tax, right?
So, we we have higher tax rates than
than some other Dominions. Um, and so
all the ad revenue in Australia for Meta
is booked through Ireland as a country
famous for Guinness Leprechaorns
uh and
and sheltering Silicon Valley's loose
change with their low corporate tax
rates.
Now, uh, in Austria, I'm glad to say you
have cleverly avoid this by charging
them a 5% uh, digital sales tax um, on
all their revenue. So, so they can't do
the same trick here in in Austria and
not pay any tax at all. Um, but
this is the irony. Meta have increased
their ad rates here in Austria compared
to other countries by 5%. So, it's not
them that's paying the tax, it's their
advertisers that paying the tax.
So, I just don't understand it. How is
it that some of the most profitable
businesses that exist today can decide
that they don't want to pay taxes?
They I mean they could well afford they
they're sitting on vast cash mountains.
It's not like they're buying their
shares back all the time. It's not like
they couldn't afford to pay the same tax
rates as other corporations, older
corporations or or you know citizens
like you and me.
Um
my my view is that if a corporation
decides that you know it doesn't want to
pay tax on a particular uh dominion
doesn't want to contribute back to the
economy that generated that wealth then
probably we shouldn't allow them to
extract wealth from that place.
Again, let me give you an Austrian
perspective to see how outrageous this
is, right? Um, Meta
uh makes about the same money in Austria
from scam adverts as criminals in
Austria make from all the illicit drugs
that are sold here
or it's about again the same as all the
counterfeit gods goods that sold in
Austria completely.
So, as far as I can see, they're
facilitating what is a huge amount of
crime, and I think that's outrageous.
Now, I I could give you many more
examples of where my anger has spilled
over into outrage about the tech
companies, about how they're bringing AI
irresponsibly into our lives. Um, but I
want to pick on another target before I
finish my talk. Um, I mean there are
many there many things other things I
could have said about the tech companies
about AI companions that undermine our
ability to connect with each other. AI
doctors that are offering dangerous
medical advice. AI nudify software that
weaponizes the the abuse of women. And
yes, I'm looking at Mr. Musk there. Uh,
or AI deep fakes that are destroying our
idea of truth. The distinction between
truth and untruth.
um threatening the cohesion of our
society, of our democracy, the
functioning of that. Um but I'm going to
turn my angry head onto, as I said, a
second target, and that's the
politicians who are letting this happen.
It doesn't have to be this way. And
again, I'll restrict myself again to
three examples. Um and there are three
examples here where I think my anger is
not transforming into outrage, but
despair. Despair that the politicians
are sitting on their hands and not doing
enough as usual.
Uh first of all, I'm desparing that that
in many places
governments are not investing enough in
the upside to justify some of the
downsides that we're going to see that
there is huge great upside. Um now
Austria is actually doing quite well in
terms of investing in research. I looked
up the numbers. 3% of GDP. You should be
proud that your government is doing
that. But only half a percent on
fundamental research. And you need to do
better. That's not enough in fundamental
research. Certainly, we're not in back
in Australia where I am that we're not
investing anything like that. We're
we're actually below the G20 average by
a long way. In fact, we're half of the
G20 average. Um there's much more
fundamental research we need to invest
in art intelligence to to build true
intelligence in machines. Um and uh
places like our universities place to be
to doing that. And if developed
countries like Austria, Canada,
Australia to commit to compete in this
AR race against, you know, the big
countries like the US and China, um
we're going to have to lift our game.
The second area in which I'm somewhat
desparing is how governments are not
regulating appropriately the harms that
I've just talked about of artificial
intelligence.
Now, the EU has led the way in terms of
regulating AI. I will give them credit.
there. The EU AI act is the first major
piece of AI regulation that exists. Um,
and there's, you know, related
regulation like the digital services
act, the buttressing what you're doing
with the EU AI act. Um, but it's very
clear over the last two years that the
appetite for regulating AI has well the
digital space as a whole, but especially
of AI has dropped significantly
um, ever since the Trump presidency,
second presidency uh, started.
Yeah. In fact, on the first day of his
second term, uh, one of the first
executive orders he signed that day was
to undo all of the regulation that Biden
had put in AI regulation that that Biden
had enacted um in his previous term.
Now, we don't actually have to um
necessarily introduce lots of new
regulation. There's actually lots of
things that we could do with existing
regulation. We could we for a while I
think we thought that you couldn't
regulate the digital space and you
shouldn't regulate the digital space. Um
you couldn't because it was somehow
different in the physical space but
that's not true. It's exist on servers
exists in data data centers. These
companies operate in particular
countries laws apply to them. The
digital space is does not have to be a
wildware. So we could apply more
forcefully many existing laws that we
have about about product liability,
about competition, about privacy um to
the tech companies. But it's clear and
we should do that. But it's clear that
there's also fresh arms. There will be
things that were not imagined when the
politicians were drawing up those rules
many years ago that could the
politicians could not have imagined. You
know, things like the deep fake nudes
that I talked about here. And indeed,
you know, is an example of what you
should do and what has been done in at
least one jurisdiction was in Australia,
um we introduced new regulation very
recently in the last couple of years,
last year indeed um to criminalize the
distribution of deep fake news was
causing significant harm especially in
certain certain um schools. Um and it
was not a criminal offense to circulate
synthetic images because they weren't of
any real people. Um, so we made it a
criminal act act so that people would
understand the severity of what they
were doing and that it could be better
policed.
And other countries are enacting or have
enacted other AI laws to protect their
citizens from deep fakes or from
manipulation, manip misin misinformation
and the other significant AI harms that
are starting to emerge. But there's many
new ones that we have only starting to
understand. Um, and I have just one
question for the politicians that I have
never seen answered well. And that
question is, did we not learn anything
from social media?
We can't simply let the tech companies
regulate themselves. They have
demonstrated convincingly that they
can't do that. And we're about to
supercharge the sorts of digital harms
that we saw with artificial
intelligence. much more powerful, much
more persuasive content that is going to
be um way more disastrous if we're not
careful than social media ever was. And
it surprises me that politicians are
actually somewhat more reluctant to to
regulate AI at the moment. Um
to take an example in Australia we've
just passed a world first social media
age ban for teenagers
that we've decided on the balance of
evidence that social media seems to be
causing harm amongst teenagers, people
under the age of 16. Um it's not a safe
space for them. Uh and that maybe we
should like alcohol and tobacco and
various other things that we decide uh
are not appropriate for young formative
minds. they need to be protected from
and so young people in Australia are now
protected from the harms of social
media.
This has been proven to be highly
popular. Almost every parent thinks it's
a fantastic idea. You can now tell your
kids you can you need to get off social
media. Even teenagers have come up to me
and said it's a fantastic idea. The
pressure is off them to be on social
media.
Uh and subsequent to that law coming
into effect at the end of last year,
there's over a dozen countries now that
are considering or enacting similar uh
age restrictions on social media. The
UK, France, Spain, Denmark, Greece,
Slovenia, Portugal, Indonesia, Ireland,
India, and as I understand here in
Austria.
Um, I'm I'm not a politician, you can
tell, but it seems to me that regulating
AI would be a real vote winner as well
as being the right thing to do.
So, that brings me to my third despair
with politicians and that's that the
political conversation is now being
dominated by big tech itself. Big tech
has embraced the idea of lobbying in
Washington uh and London and now it
seems also in Brussels and CRA. Uh in
the last Australian election, the tech
sector has donated more to the political
parties than any other sector combined.
Uh the tech sector here in Europe uh is
also one of the biggest spenders in
Brussels.
That should tell us something. they're
not spending that money um other than
because they think it's going to be
helpful for their cause.
I I want to give you just one example,
one personal example of how politicians
are unfortunately starting to surrender
to the persuasive messaging of big tech.
In February last year, 2024, the
Australian government rather wisely
decided that they needed some
independent expert advice on art of
intelligence. Um there are huge
financial incentives for the tech
industry to do the wrong thing to move
fast and break things like the mental
health of our young people. So they
thought some independent technical
advice not coming directly from the tech
industry might be a good idea. So they
set up what was called at the time a
temporary AI expert group of academics,
lawyers and other people um on the
fringe of tech um to offer that them
that advice. Now, full disclosure, I was
one of the 12 people who was was asked
um to be one of those independent
experts to the Australian government,
offering them advice um unpaid advice as
usual about how Australia should profit
from the race and how we should avoid
the harms. Um unfortunately, the
temporary in our name, the temporary AI
expert group proved to be entirely
precient because we're now being closed
down. All the problems have been solved
obviously.
um despite them promising us they would
make us permanent despite asking us to
apply for a permanent group and the um
uh the government quietly revealed in
December that they wouldn't be
continuing the group and indeed that
they wouldn't be making any new
regulation for AI
which again that's bucking so many
international trends. The UN, I'm glad
to say, has just appointed 40 experts in
its independent um international
scientific panel, bit like the ICC, uh
the climate panel on art intelligence.
US has a national art of intelligence
advisory committee which reports
directly to the president. Japan's PM
set up the artificial intelligence
technology strategy council. South Korea
has a national AI strategy committee
with 50 commissioners, many independent
experts. Ireland has an independent AI
advisory council and Austria I
understand has for many years had an 11
member barat for countis intelligence an
AI advisory board as well as an AI
policy forum and an AI stakeholder forum
so good on you Austria but you know I
have just one question for politicians
that again I've not seen answered very
well wouldn't it be advisable to be
seeking independent advice rather than
listening to the lobbying of big tech
because there's huge financial pressure
as I said not to do the right thing.
Again, it surprises me that they're not
wanting to listen more to outside
independent voices. Why wouldn't they
want some independent advice given at no
cost as usual? Um,
I can assure you that I however that I
will continue to offer fearlessly my
advice whether the politicians want it
or not. Now, now I'm actually going to
offer the advice not in private as I
was, but in public as I am doing today.
I'm sure they're going to find this more
uncomfortable than I'm going to find
this uncomfortable. I'm used to giving
lectures in which no one's listening.
Um, and I'm going to end before there's
plenty time I hope for questions. Um,
with one final emotion, and it's not
anger, it's fear.
I fear, as I intimated, that we're going
to be repeating the mistakes of social
media.
Social media should have been a wakeup
call about the harms of unregulated tech
that we're about to supercharge with, as
I said, much more powerful and
persuasive technology than social media
ever was. What I fear most is that I'm
going to be back here in three or four
years time saying, "We tried to warn
you, but another generation of young
people has been sacrificed to the
profits of big tech."
I have so much more anger and outrage I
could have shared with you. I could have
talked about the environmental harms of
AI.
Uh uh I could have talked about concerns
about AI powered workplace surveillance.
I could have talked about the impact of
AI in jobs especially graduate level
jobs.
In fact, the most common question I get
today um is what can my children study
to make them AI safe?
And I could have talked about how AI is
transforming warfare as we see sadly in
Iran and Ukraine and other hotspots of
the world. In fact, I was um privileged
to be at the UN in Geneva a few weeks
ago talking again about killer robots.
Now, that's a sentence back 50 years ago
the young me would never have imagined
to have said. There's so much more I
could have talked about about how some,
but thankfully not all of my young
dreams are turning into uh nightmares,
but I'm going to end not with my own
words, but some of the final words of
how 9000 from that film 50 years ago,
2001, a space odyssey. They're the words
that began my journey here today to the
technical university of Vienna. And
they're the words the AI said when the
humans started to ask difficult
questions.
Hal,
I can see you're really upset about
this. I honestly think you ought to sit
down calmly, take a stress pill, and
think things over. I've still got the
greatest enthusiasm and confidence in
the mission, and I want to help you.
Thank you.
Thanks, Toby.
Uh questions from the audience.
>> Uh yeah, fascinating presentation. I see
there's plenty to worry about. Um
>> Oh, well, you were obviously paying
attention.
>> Yeah. Yes. At least one was.
>> I was.
>> Okay. Uh and I just finished a book
which I'm going to assume you're
familiar with called If Anyone Builds
It, Everyone Dies. Uh, I just finished
it this morning, so I'm hoping you'll
give me some reasons to sleep well
tonight in spite of what I just heard
now.
>> Yes.
>> Uh, as the book states, we should stop
further development of AI to the
threshold past the threshold of what is
superhuman intelligence. So, actually
have a twofold question. One, how real
is this danger? Is this overexaggerated
or is there still a probability of what
they're foretelling of super AI
destroying humanity and taking over the
planet and the world? Um, and if there
is a yes to that or a no to that, how do
we know when we reach super human
intelligence? Uh I I don't think there's
a a gauge on the side of the computer
saying you are now approaching
superhuman intelligence. Please stop.
>> Yeah. Well, fasc fascinating question.
There's a there's a long answer to the
question which is I've written a whole
book which is coming out in September to
answer that question. I'm going to have
to give you the short answer.
>> Executive summary. the executive summary
which is um intelligent people
overestimate the importance of
intelligence
right so
the university you're surrounded by some
of the most intelligent people and yet
the problem in the problem is not
intelligence the problem is power is
that what people do with their
intelligence is the power the power that
they have and it's
if if intelligence was the important
thing you should hang around
universities because they would be the
center of changing the world. They're
not. We tried we we struggle to be to to
to actually make any difference to the
world because power resides elsewhere.
The and that we've set up a complex set
of institutions to limit the excesses of
power. You can't go off and do what? You
can't turn the world into paperclipip
factories because we have environmental
planning laws. We have environmental
activists who will stand in front of
your bulldozers who will prevent you
from turning the world into paperclip
factories. So it's
in intelligence is often not the thing
that's holding us back and it's not the
thing that will allow you to destroy the
planet. As an example, um the problem
with you know the greatest existential
crisis that faces humanity today is the
climate emergency. I think it's it's
pretty clear that that is something
that's going to have the greatest harm
to the greatest number of people is the
way that we're changing the climate of
the planet. And we know it's not
intelligence that's that's stopping us
fixing the climate. It's human stupidity
in human politics and money. Yes. And
that we haven't fixed that. So it's not
that we need a a greater AI intelligence
super intelligence that's going to come
up with a better plan. We already 20
years ago we already knew what the plan
was, right? Um but we haven't been able
to execute that plan. So I'm actually
much less worried about artificial super
intelligence than I am about artificial
stupid intelligence. the fact that we
will hand over responsibility as we are
doing to artificial intelligence to
decide targeting decisions in Iran or um
to decide welfare payments in uh the
United States or whatever it is that
isn't actually smart enough or doesn't
have the capability to be making those
decisions. So I I'm pretty relaxed about
the threat that um artificial super
intelligence has. I actually would like
more intelligence on the planet to help
us solve some of the wicked problems.
>> Thank you.
>> Thank you very much, Toby. This just
reminds me about the paper I think you
did maybe 10 years ago or so. The
singularity is not near.
>> Why? Why the singularity may never be
near.
>> Yeah, exactly. So maybe was to to have a
question on this. So I I now recognize
you didn't change your mind much
although this was before church GBD and
so but on the other hand if we if we
listen to the to the big tech and and
how they they they now I mean all this
hype and this many many investments I
mean they are talking about hi they
promise hi
>> yes
>> um will they fail
>> uh no I I think it would be conceited to
think that we couldn't build
intelligence that matched and probably
exceeded human intelligence Why why
would human intelligence be the supreum,
the limit of intelligence? We're an
evolutionary accident. We happen to be
as smart as we were. We were one of the
smartest things around when we when we
um you know came off the savannah and
we've used that intelligence to good
effect. Everything in this room is the
product of human intelligence. The
quality of lives that we live, life
expectancy has doubled here in Austria
since the industrial revolution because
of the things that we have invented as
humans. So intelligence has been a a
great gift. But our superpower wasn't
our intelligence. Our superpower was our
society. Our ability to come together
and solve problems collectively. We
already have super intelligence.
It already exists in corporations and
governments. No one knows how to build
an iPhone. No one knows how to build,
you know, the people who work for Apple
and the subsidiaries between them. No
one knows how to build a nuclear power
station. people who work for
Westinghouse collectively have that
knowledge. So we already have these
organizations that are smarter than
their parts that can do things that we
couldn't do individually. All of us all
of us put on a desert island on our own
would quickly starve. Well, I certainly
would. I imagine most of you would as
well. We've lost most of those skills
that we had of looking after ourselves
because we have society. That means that
you know we can be we can do fanciful
things like being AI researchers and
other people manage to make food for us
so that we don't starve. Um and so it's
that so so it and the artificial the
super intelligence that exists in
corporations isn't destroying the
planet. I mean it's not it's not
perfectly aligned. You know 70 companies
are responsible for for most of the
global emissions. we could need to get
those better aligned with human
flourishing. But but I'm not I don't
think that's an existential threat. I
don't think shell oil is ultimately
going to destroy all of humanity. It may
make life a bit more painful for many of
us, but I don't think it's an
existential risk. I think it's something
that we can manage. As we've managed, we
built this complex society, all these
different institutions that are
competing against each other, providing
friction to stop anyone having too much
power. I must admit it's gone a bit
wrong in the United States recently. But
I'm confident in a few years time we'll
have fixed that and the US might get
back to being a more reasonable place.
But I don't think the I don't think you
know societyy's going to collapse. I
think it's going to maybe it's going to
be more painful than it should be. And
the the the sad thing is that every
generation inherited a better life than
the previous generation. And we have
spectacularly turned that one around and
it looks almost certain that our
children will have a worse life. and
we're responsible for that failing.
>> So in very short it's not the technology
but how it's structured and how the
industry is.
>> Yes. Yeah. Yeah. Technologies are just a
tall but it's how we you have to think
about you know how do we revitalize our
society our institutions the US is a
wonderful example of how the
institutions are failing it and we need
to think about how do we refresh them?
How do we how do we build them so that
we can continue to flourish as as a as
humanity?
>> Did you raise your hand?
>> I'm bringing
>> just behind you.
>> Thank you so much. Uh you mentioned uh
the EUI act um and the struggle for
power in the AI development and I feel
like as the UI act is definitely being
watered down Yes. in now by countries.
>> Yes. Um, the new paradigm that's on
vogue is digital sovereignity. Um, at
least within Europe, and I know that
this is on top of the agenda in Austria
as well. Um, and I'm wondering what kind
of effect that could have on AI
development, if you see any potential at
all in in this.
>> Yeah. Um,
I think we're in a unfortunate moment in
technical history, which is that at the
moment it's who can outspend who who who
got the biggest models, who
it's a it's a scaling race at the
moment. But that's actually not what
ask
us to get there quicker. Um, but I'm to
get there quicker. Um, but I'm pretty
confident that's where we will end up.
>> Further questions?
Oh yeah. Okay. Uh yes. I I'm not an um
technic guy. I'm more into in philosophy
and I'm wondering uh for me the main
problem is ourselves, our attitude to
life,
the whole system of capitalism, the
whole system of competitive science. So
this is the main reason for our
problems. Not the artificial
intelligence, not the technical gadgets
that we are inventing but uh our own
attitude to life. So we should have
uh afraid of it because we maybe really
unconscious know that we are mean
persons that we will misuse it.
Everything that mankind has invented we
will misuse it in a way. Yeah.
>> And we will will try everything because
we are very uh no
>> curious
>> curious people. So if we are curious and
mean people this is a bad combination.
And uh all the talk about how we can uh
regulate AI, we is is going to fail