Predictions of humanity’s demise at the hands of intellectually
superior software have fueled calls for curbs on the development of artificial
intelligence.
اضافة اعلان
AI groundbreaker Geoffrey Hinton recently made waves when he
warned that his creation could one day
threaten humanity – especially if AI systems learn to write and run their own
code (killer robots, anyone?). An
open letter signed
by Tesla’s Elon Musk and Apple’s Steve Wozniak warns that advanced AI could
eventually render human minds obsolete.
If all of this is true, how do we, the potential collateral damage
in this coming conflict, prepare?
‘The Art of War’
In the “Art of War,” Chinese military strategist and philosopher
Sun Tzu observed that to win in battle one must know themselves as well as
their enemy. But in the coming war with AI, it may be just as important to
understand what AI knows about us – and how to influence the storyline.
Much of AI’s power comes from what we feed it. Social media giants
like Facebook, Instagram, and Twitter use AI, powered by our clicks, likes, and
views, to
produce a picture of our habits, buying preferences, and interests. AI fuels what we see online,
but its vision is fed by our own actions. Organizations like the
Panoptykon Foundation,
in Poland, and the
Association for the Advancement of Artificial Intelligence, are advocating for laws that protect people
from intrusive AI and the societal implications it carries.
Curious to understand how well my “enemy” knows me, I spent a few days chatting with AI about, well, myself. Our conversation began on Facebook.
To be sure, there are
plenty of beneficial applications. AI is revolutionizing cancer research, weather forecasting,
materials development – even
mail delivery. From smart fridges to traffic lights, AI is already part of our
lives in ways we rarely see. And yet, when the very people who created the
technology tell us to take heed, we should listen.
How well my “enemy” knows me
Curious to understand how well my “enemy” knows me, I spent a few
days chatting with AI about, well, myself. Our conversation began on Facebook.
Facebook collects huge amounts of data on its users and has
come under fire for doing so. Meta, which owns the platform, recently agreed to
pay
$725 million for sharing user data without permission. Fortunately, the platform also makes
it
easy to obtain the data itstores.
With a few clicks, Facebook will deliver users a history of every reaction,
comment, message sent, and search performed. It even coughs up a compilation of
topics that its AI has determined are a users’ “ads interests,” based on
activity. Better than any photograph, these keywords are essentially a digital
file of who Facebook thinks a user is – the consumer behind the keys
that it can monetize.
What I apparently like
With these data points – topics like “music festivals” and
“snowboarding” – I turned to the AI chatbot ChatGPT to help make sense of my
Facebook self. Plugging the 105 ad topics generated by the platform into the
search bar, I asked
ChatGPT to tell me who “I” was. The result was like a fuzzy
photo, with the clearest pixels at the margins.
To be fair, ChatGPT surmised correctly that I am a journalist, had visited Walt Disney World (when I was eight), and enjoyed running marathons, including the London Marathon (not sure “enjoy” is the verb I’d choose, but points for specificity). But it failed elsewhere: I don’t play tennis, am not a fan of ice hockey, never pursued a legal education, and don’t have a Grateful Dead tattoo (yet). All in all, ChatGPT was about 50 percent accurate in its digital depiction.
Apparently, I live “in Philadelphia, Pennsylvania (wrong), and
liked exploring the world of arts and music (depends on the artists).” I’m an
avid basketball fan (nope – soccer and baseball are my sports) and enjoy
listening to BBC Radio (I’m more of an NPR guy). I even have a “few tattoos
from Ink Master and Inked magazine” (wrong on both; hate needles).
To be fair, ChatGPT surmised correctly that I am a journalist, had
visited Walt Disney World (when I was eight), and enjoyed running marathons,
including the London Marathon (not sure “enjoy” is the verb I’d choose, but
points for specificity). But it failed elsewhere: I don’t play tennis, am not a
fan of ice hockey, never pursued a legal education, and don’t have a Grateful
Dead tattoo (yet). All in all, ChatGPT was about 50 percent accurate in its
digital depiction.
I repeated the exercise for
Instagram, yielding similarly hazy
results.
Finally, when I asked ChatGPT directly, “What do you know about
Greg C Bruno, the journalist?” not even this was convincing. While the bot said
I was a journalist (which I’d told it), it incorrectly identified where I’d
been published (yes to The New York Times, no to The Wall Street Journal) and
the places I’d reported from (
Afghanistan, no; Tunisia, yes).
Then again, the compliment might be a ruse. As Sun Tzu foresaw, “When envoys are sent with compliments in their mouths, it is a sign that the enemy wishes for a truce.” In our fight for humanity’s future amid software that’s only expected to get more powerful, perhaps a truce is the best we can hope for.
ChatGPT didn’t even get the title or topic of my book correct,
which,
according to Google, is referenced more than 4,000 times (not a best
seller, but not exactly hiding, either). Instead of “Blessings from Beijing:
Inside China’s Soft-Power War on Tibet,” I’m apparently the author of
“Blessings from Baghdad: How a Secret American Mission Saved the Allies in World
War II," which tells the “little-known story of a covert American
operation that helped turn the tide of the war in the Middle East.” That book
doesn’t even exist – the AI made it up.
There is one ChatGPT “fact” that I won’t quibble with: “
Greg C Bruno is an experienced journalist and writer who has contributed significantly
to the field of international journalism.” I'm flattered.
Then again, the compliment might be a ruse. As
Sun Tzu foresaw,
“When envoys are sent with compliments in their mouths, it is a sign that the
enemy wishes for a truce.” In our fight for humanity’s future amid
software that’s only expected to get more powerful, perhaps a truce is the best
we can hope for.
Greg C Bruno is the author of “Blessings from Beijing: Inside
China’s Soft-Power War on Tibet.” He was a term member of the Council on
Foreign Relations in New York and is a former opinion editor at The National in
Abu Dhabi and Project Syndicate in Prague.
Read more Opinion and Analysis
Jordan News
.jpg)
.jpg)
