Стэнфордын их сургуулийн төгсөгч Такер Брайант хиймэл оюун ухаан мэт дүр эсгэсэн ChatTJB үйлчилгээгээрээ дамжуулан хэрэглэгчдийн асуултад өөрийн биеэр хариулж байна.
ChatTJB нь хиймэл оюун ухаан бус, харин Такер Брайант хэмээх хувь хүн хэрэглэгчдийн асуултыг хүлээн авч, өөрөө биечлэн боловсруулж хариулдаг үйлчилгээ юм. Төслийг санаачлагч Брайант уг ажлыг эхлүүлснээс хойш өдөрт 1000 гаруй удаа хэрэглэгчдийн хоолны сонголт, гэрийн засал чимэглэл, харилцааны зөвлөгөө зэрэг өргөн хүрээний асуултад гар аргаар хариулж байна. Тэрээр хиймэл оюун ухаанд хэт найдах нь хүний бүтээлч сэтгэлгээ болон шүүмжлэлт сэтгэлгээг сулруулдаг гэсэн санааг илэрхийлэхийн тулд энэхүү сатири буюу шог шинжтэй төслийг хэрэгжүүлж эхэлжээ.
Брайант өөрийн төсөлдөө “Average Individual” буюу дундаж хувь хүн гэсэн тодотголыг ашиглан хиймэл оюун ухааны зах зээлийн сурталчилгаанд шүүмжлэлтэй ханддаг. Хэдийгээр эхэндээ зөвхөн өөрөө хариулдаг байсан ч, одоо тэрээр үйлчилгээний аюулгүй байдлыг хангах үүднээс хяналтын хэрэгслүүдийг туршиж, зөвшөөрөгдсөн сайн дурын ажилтнуудыг татан оролцуулж эхлээд байна.
Хэрэглэгчид хиймэл оюун ухаантай ярьж байна гэж андуурч хувийн мэдээллээ алдах эрсдэлтэй байгаа тул Брайант үйл явцыг удаашруулж, болгоомжтой хандаж байгаагаа илэрхийлжээ. Тэрээр хиймэл оюун ухааныг бүрэн эсэргүүцдэггүй бөгөөд технологийн дэвшлийг үгүйсгэхгүй ч зарим тохиолдолд жинхэнэ хүнээс зөвлөгөө авах нь илүү үр дүнтэй байдаг гэж үздэгээ хуваалцсан байна.
Дэлгэрэнгүйг эх сурвалжаас харах
↓Эх сурвалжийг нээх ↓
ChatTJB, according to its Crunchbase profile, is a service that “provides a personal conversational response where an individual directly receives, processes, and writes answers to user inquiries”. That’s technically true, but not in the way you think… because it’s just sorta some guy.
Futurism got to sit down with Tucker Bryant, the man behind ChatTJB, and as you might be able to expect, he admits, “I tend to overcommit to bits.”
If you’re wondering how far he’s willing to commit to the bit, Bryant rented out a billboard for $6,000, which says ChatTJB is the “leading chat interface powered by AI”. AI in this case stands for ‘Average individual”.
He has reportedly been personally writing responses to users over 1000 times a day since starting the project, answering questions on what to eat for dinner, what colour to paint their homes, and he has even given relationship advice.
As Futurism notes, “Like a chatbot, he can also generate images upon request— in the form of graphics that he hastily sketches, of course.”
Bryant is a Stanford graduate and a former Google employee who claims the idea came off the back of his partner telling him about “cognitive surrender”. Effectively, it’s argued that those who use AI chatbots are likely to blindly trust the information given to them, even when the information is incorrect.
Bryant says, “We often, or increasingly, trust LLM outputs so easily that we stop bringing our own creativity or critical thinking into the responses we receive, and over time that weakens our ability to be fully-functioning, sovereign human beings in the world.”
Though Bryant was expecting some trolls and acknowledges ChatTJB is a bit of a troll itself, he says it has been “really sweet” to chat with real people and even noticed some users asking more serious questions. One user reportedly said, “I’m on the first night of my honeymoon and I don’t feel relaxed, is that okay?” and Bryant “switched out of super-casual-mode into supportive-friend-mode.”
Though that does raise some concerns. If users are talking to ChatTJB and are under the impression they are talking to an AI anonymously, they could reveal personal information in the process to another human being. That could become more of a concern as ChatTJB has seen interest from users online who want to become the bot on ChatTJB.
However, Bryant is taking it slowly and figuring out how to get volunteers involved. “This was just intended to be an art project. I think it’s beautiful that people want to do something less ridiculous with it than what I had planned.”
He continues, “I also don’t want people to get hurt, so I’d rather people wait a long time for their prompts than open the floodgates to anyone and everyone. So, a long preamble to saying: I’m starting to let some approved volunteers try answering questions while testing out some moderation tools.”
Bryant argues that ChatTJB is a satire of AI and the culture surrounding AI advertising. “That’s the highbrow answer. But the lowbrow answer is just as authentic, which is that I also thought it would be funny.”
Bryant is not explicitly anti-AI, though. He says on his Instagram, “AI can do some really cool things, but sometimes all you really need is to talk with an average individual.”
