A painting featured in the film “The Matrix Resurrections.” A piano playing computer-generated music based on the activities of a bacteria colony. A small army of Spam cans with mechanical hands, all collaboratively pecking at keys to type a story composed by artificial intelligence.
At times funny, at times chilling, the exhibits at the Misalignment Museum’s grand opening earned curious glances Friday evening as passersby peered through the large windows of the art installation at 201 Guerrero St. near 14th Street.
The museum, which will be open through May 1, aims to deepen public understanding of the potentially sweeping consequences of supercharged AI. Following a year marked by leaps forward in AI’s capabilities, attendees, as well as the museum’s director/curator, Audrey Kim, expressed a mixture of excitement for new technologies and fears that humanity may be unprepared for what comes next.
Tech employee Neal Grantham said he is “excited about where we are right now,” and that he’s had fun interacting with the popular text generator ChatGPT, which launched in November.
“I’m OK right now,” he said. “Ask me in five years.”
By contrast, Tatiana, another tech employee, who declined to give her last name, said she has “a bit of a nihilist perspective” on the technology. She worries that sufficiently advanced AI will have devastating implications for the world as we know it.
“I just think that, ultimately, our species is just the incubator for the next intelligence, which is AI,” she said with a partly joking smile.
Kim, who works for the art production group Eurypheus, curated the show over four months, supported by funding from an anonymous donor. She imagines a post-apocalyptic world in which AI seizes control of most of the globe in a bid to gain more computational resources. After slaughtering almost everyone alive, the AI realizes the error of its ways and creates the Misalignment Museum as a memorial to civilization.
“Sorry for killing most of humanity,” reads a caption that is visible from the street.
Kim, who got interested in AI safety while working on self-driving cars at Cruise, isn’t sure whether such a horrific chain of events could really happen. But with the explosion of computer-made art and literature this past year, thanks to ChatGPT and the image-creating DALL-E (not to mention Microsoft Bing’s sometimes-confrontational chatbot that launched last month) she believes conversations about AI’s potential are more important than ever.
“I think there being something to look at that is tangible can increase understanding and knowledge,” Kim said.
Among the museum’s most striking exhibits is “Paperclip Embrace,” a sculpture composed of roughly 15,000 paper clips depicting two people embracing. This is a reference to “the paper clip maximizer,” a thought experiment in which programmers instruct AI to produce as many paper clips as possible.
With no sense of morality, the AI in the thought experiment creates paper clips at all costs. Because humans could potentially destroy the AI, negating its ability to make paper clips — and, since human beings are wasting atoms that could be turned into paper clips — the AI decides to kill all humans.
For Kim, this thought experiment illustrates the importance of caution and careful planning as technology advances.
“I think it’s highly unlikely that AI will extract my atoms to turn me into paper clips,” she said, laughing. “However, I do think that there are a lot of destructive outcomes that could happen with this technology.”
Kim is especially worried about the harmful possibilities of artificial general intelligence (AGI), which refers to an AI that could understand or learn any task that a human being can. She worries that the lack of a “control valve” on such intelligence really could destabilize civilization or even destroy humanity.
Hence the name of the Misalignment Museum: a “misaligned” AGI would be one with an underdeveloped knowledge and acceptance of ethics.
The caption hanging above the entrance to the museum’s lower floor reads, “Abandon all hope, ye who enter here,” a reference to the inscription on the gates of hell in Dante Alighieri’s “Inferno.” Kim doesn’t think people should abandon all hope. But she does think people should take technology’s possibilities seriously.
Kim sees these possibilities as frightening, but also potentially positive. “Like, the future is not set,” she said.
Kim hopes that the Misalignment Museum will encourage people in the tech industry to think more carefully about the implications of their work, realizing the impact it could have on society and future generations.
The exhibit isn’t just for techies, though. Kim hopes that it encourages visitors from all walks of life to realize the role that technology is already playing in their lives and be more thoughtful about issues like digital privacy.
“I want people to be more aware, and to not just go quietly into this good night,” she said.
A successful launch
Hillary Clark, one of two artists responsible for the paper clip sculpture, is on board. She sees art-generating software as a powerful tool that artists are just now beginning to explore.
“We have this opportunity right now to create a new dialogue about what tech is,” she said.
Visitor Paul McKellar, who has a master’s degree in computer engineering, agreed. He referred to AI as “the technology of our time,” but said it’s difficult for people to wrap their heads around it when so many of its most extreme implications would require a huge jump in its abilities.
“I think it’s reasonable to be worried. I think it’s hard to be worried, because it’s very far away,” he said.
Regardless of background, visitors laughed and looked with interest at many of the museum’s quirkier installations. These include a kneeler at which visitors can speak to an AI that responds with the “voice of God,” as well as a pair of roving Roomba vacuum cleaners outfitted with brooms.
Kim said she is surprised by how much support the Misalignment Museum has already received. One hundred tickets for the grand opening sold out within two days, and the venture firm Factorial Funds committed $250,000 to the museum Friday.
She hopes to raise $2.5 million for the exhibit by May 1, which would allow the museum to get a larger, permanent location.
Kim said Saturday afternoon that she is unsure what, exactly, is drawing so many people in, but the museum seems to be “resonating” with people in the Bay.
“Something is going on,” she said. “It’s like, ‘All right. We’re gonna all figure this out together.’”
The museum is open to the public Thursdays and Fridays from 4 to 7 p.m. and Saturdays from 2 to 5 p.m.
Maybe don’t venture into a space that has…..“Abandon all hope, ye who enter here” over the front door?? Just a quick thought from an old crow! 🙂
Thank you for this article on such an important phenomenon that has been put forward in a prescient, thoughtful, and good humored way.
On March 5, h. brown wrote :
Please forgive and/or indulge my nitpick. Chinggis (Genghis) Khan did not go into or near Ukraine. While administering his empire in Central Asia, he ordered his son Juchi and 2 prominent commanders Jebe and Subotai on a reconnaissance in force from what is today northern Iran (transliterated spellings of all 3 names vary) through the Caucasus Mountains and north into what is today south Russia and Ukraine. From 1220-23, in what was called The Great Raid, they raided and explored the vast region, then returned through what is now Kazakhstan to Central Asia.
I prefer this to SFMOMA.
I am mostly concerned at this time that AI will take my job and many others.
There are many who are just too old to start training in a totally new job, though too young and too poor to retire. I have already lost parts of my job that a bot does instead. I wonder how customers and patients will feel talking to bots vs humans?
“I wonder how customers and patients will feel talking to bots vs humans?”
We already know that where Bots are in play today:
Bot (More blather)
Bot: (Even more blather)
Can it create a Panel to discuss the Ukraine War for me ?
I’m thinking of Genghis Khan who fought there.
And, Stalin, of course who knew the landscape well.
Napoleon would have to be sitting there to scrutinize modern artillery.
Hitler has to have a seat
Think anyone would watch their discussion ?
For Moderator I’d use Henry Kissinger.
Go Niners !!
No. The moderator should be Isaac Babel, or an AI facsimile.
I’m serious about this.
I stumbled across a battle plan for that region that is 2,000 years old.
All of these invaders (all after the Wheat fields) had plans that are now published.
It would seem easy for the right Joe Rogan guest to bring them all together.
What was it that Santayana said ?
“He who does not read history is bound to repeat it.”
No, not that one; the other one …
“Only the dead have seen the end of war.”
Yeah, that one.
Will AI change that ?
Go Niners !!