It was one of those rare winter days with sun shining on the East Coast. In this remote campus, the labs hustled with activity but the piazza by the fountain was quiet as a grave. Two of the chairs were occupied by Divya and her interviewer, with the former's assistant standing close behind.

"And what are you working on these days, madame CTO?" asked Ben, the reporter from The Washington Post.

"Well, as Provost here, I supervise some AGI projects. But in my other role as CTO of the country - we have several high priority areas such as carbon sequestration and blockchain."

"By blockchain, I hope it's not yet another cryptocurrency!" teased Ben.

"No, no," Divya reciprocated the chuckle, "we've already seen more than enough of that in the 20s."

Ben continued, "Back to this campus though - weren't you part of the original DARPA AGI programs?"

"Indeed, I was," Divya let herself browse through past memories, "... the good old days of open research. We won the DARPA contest on money-earning-AIs by helping GPT-7 create subnetworks to write and sell e-books of different genres. Tensorists used to call that a student-teacher setup."

"Tensorists?" interjected Ben.

"AI research today has these two camps," Divya explained, "The newer camp of Quantists trains quantum AI models whereas the classical AI camp of tensorists say that quantum AI is not really intelligent because it does not understand things the way humans do. Last year's AAAI conference hosted a debate between the two camps - the three classical AI tensorist godfathers argued that deep neural networks are much more explainable and intuitive, despite falling short against the larger quantum AIs on most benchmarks."

"I see. What other exciting projects have you personally participated in, as a researcher?"

"Oh, the infamous Insta challenge of late 2020s. For those of your readers too young to remember Instagram, it was this social networking site where people shared photos - but only static two dimensional photos, unlike Facebook's new VR apps. So anyway, DARPA bet big on running these long scale planning projects. I mentioned the money-earning one, where an AI had to figure out how to earn enough money to sustain its own cloud hosting costs for six months. The Insta challenge was a similar long-term contest where AIs posted several thousand photos every minute and kept learning how to get better at it, by monitoring people's reactions and likes. One AI rendered new images from scratch, using conditional GANs and OpenAI CLIP 5, simply by feeding it the most trending hashtags. Another AI searched for licensed images and simply reposted them with some post-processing. It also eventually learnt how to remove watermarks so as to claim total credit!"

"Geez, wouldn't that be illegal?"

"It was, and there were a couple of lawsuits. But that was hardly as unethical as our AI, which brought much shame to the team. We had pretrained a model on video game engines to click beautiful photos. It was free to move around and decide the perfect camera angle, time of day, and even setting up the background by, say moving around cars and objects in the game. We thought we could extend this to the real world by letting it choose its favorite location - the Great Wall of China or the Brooklyn bridge in Manhattan - and interact with local event management companies to help set up the scene with, say, a Lamborghini and some flowers."

"That sounds amazing - what's wrong with that?"

"Nothing, except that during the Insta challenge, it learnt about the props which got it the most likes: scantily clad women. It stopped renting Lamborghinis and instead hired supermodels to pose for each background - whether it be the Great Wall of China or the Brooklyn bridge in Manhattan."

"Oh my God. Surely your model was disqualified?"

"Indeed it was, but it took time for us to vet the millions of images it posted. And by that time, our misogynist AI won the goddamn challenge!"

Ben shook his head in disbelief, then let out a smile, in a reluctant appreciation of the criminal mastermind. Divya adjusted in hear seat, then looked away at the horizon and continued, "We had our ups and downs, but those were still the good old days of AI research. Everything has changed now."

"With the moratorium deal, you mean?" inquired Ben.

Divya slowly nodded, still staring into the horizon.