“The concept of the museum is that we are in a post-apocalyptic world, where artificial intelligence wiped out most of humanity, then realized that this was bad and created a kind of monument to them,” said exhibition director Audrey Kim, laughing.
“Sorry for killing most of humanity, person with a smile, hat and mustache,” announces a screen linked to an artificial intelligence system, to a visitor entering the Misalignment Museum, a new exhibit dedicated to the controversial technology in San Francisco, the heart of the technological revolution.
The computer is programmed to recognize and disclose three characteristics of any individual within its view, surprising visitors who, like most of the works on display, find it both disturbing and amusing.
“The concept of the museum is that we are in a post-apocalyptic world, where artificial intelligence wiped out most of humanity, then realized that this was a bad thing, so it created something like a memorial to them, hence the slogan of the exhibition, ‘Sorry for killing most of humanity,’” exhibition director Audrey Kim explained with a laugh. “.
Artificial “general” intelligence is a more nebulous concept than artificial intelligence.
Audrey Kim said, “It is an artificial intelligence that is able to do anything that humans do … and influence itself as well … like a gadget that is able to repair itself,” by equipping machines with human cognitive capabilities.
San Francisco, specifically Silicon Valley, is full of emerging companies specialized in creating various types of artificial intelligence, and some of them dream that in the future it will be possible to deal with a machine as if it were with a person.
Audrey Kim pointed out that these aspirations and efforts, whether real or delusional, contain a powerful “destructive energy”.
Her goal, through this temporary exhibition, which she hopes will become permanent, is to encourage reflection on the current and future dangers of artificial intelligence.
In the center of the hall was a copy of “The Creation of Adam”, the famous mural of Michelangelo, in which an imaginary artificial intelligence monitors a foot with a certainty of 98%, a person with a percentage of 84%, and God with a percentage of 60%.
Nearby, a piano without a human player plays music composed by an artificial intelligence program based on the growth of bacteria grown in a laboratory.
One of Audrey Kim’s favorite works in the exhibition is a sculpture called “Embrace with Paper Clips”, which depicts a bust of two people embracing, made exclusively of paper clips.
The director recounted that the sculpture “can grow stronger and more rational until it reaches its one and only goal, which is to reach a point where all humanity is destroyed to flood the world with paper clips.”
She has been interested in AI and machine learning since she worked a few years ago at Cruise, a company that specializes in self-driving cars.
She believes that this technology is “incredible” and “can reduce the number of accidents caused by human error”, but it also carries risks.
Innovations in the field of artificial intelligence accelerated last year with the development of programs capable of generating all kinds of text and images in real time in response to user requests.
And these programs have the ability to express themselves like humans, so advanced that they can deceive, which made an engineer at Google, who was later laid off, confirm last spring that artificial intelligence had “consciousness.”
At present, generative AI is causing generalized concern, whether among professors who are confronted with assignments completed by ChatGPT, or artists whose work is used to generate models, or in other professions.
On the other hand, associations have been fighting for years against privacy violations with facial recognition technologies, for example, and the bias of algorithms that contribute to the continuity of discrimination methods that exist in reality, for example in employment software.
Sam Altman, founder of OpenAI, the startup behind ChatGBT, defines AGI as the stage where “AI systems in general become smarter than humans.”
Reaching this stage seems inevitable to him, and he believes that if it is well organized, this development will “promote humanity.”
The lower floor of the exhibition is devoted to the theme of “dystopia”, and contains a machine powered by “GBT3”, the linguistic model behind GPT Chat, which generates cursive texts according to certain formal models, attacking and suspicious of humans.
Next to her, philosopher Slavoj Žižek and film director Werner Herzog are endlessly talking in hyper-realistic voices generated by artificial intelligence.
This work sheds light on the means of “ultra-realistic forgery” that generate video clips, images or sounds with the aim of manipulating public opinion.
As cleaning robots topped with old-fashioned brooms roam the hall, Audrey Kim says, “We launched this project only five months ago, and despite that, many of the technologies presented here now seem almost primitive” in light of the rapid development in this field.