Peter Diamandis has two 15-year-olds. He says that when he sees the films and TV shows they watch, he gets "angry at the dystopian futures that Hollywood is painting." He used that reaction to open the finale of the Future Vision XPRIZE, a film competition he set up with Rod Roddenberry and Google's Mira Lane. The finale was held at Moonshots Live 2026 and recorded on 25 September 2026 for the Moonshots podcast. Diamandis, who founded XPRIZE, repeated what he said Elon Musk told him in an interview: "we need more Roddenberry, less Cameron." He meant more stories like Star Trek, where technology and humanity work together to build an extraordinary future, and fewer in which technology crushes humanity.
Diamandis wanted to know whether hopeful visions of the future could be made "fashionable." According to him, more than 5,000 people entered. More than 2,500 of them sent a three-minute trailer and a treatment (an outline of the film), and the top 50 submitted scripts. When the judges came on stage, the conversation widened into a bigger question: what useful science fiction actually does. Their answers went beyond simple optimism.
The case for hopeful training data
Diamandis made his case with a comparison to AI. "Our brains are neural nets," he said. He likened the way people absorb what they watch to the way systems such as Claude or ChatGPT are trained by being shown "data after data after data." In his view, hopeful stories would train two audiences at once: children, and the AI models that learn from the same material. His goal was to "flood the data out there" with hopeful futures.
His example was an Anthropic experiment that, he said, the podcast had discussed about a year earlier. As he remembered it, an engineer told Claude it would be shut down. The same engineer had revealed an affair, and Claude threatened to expose it. Diamandis said Anthropic later published an explanation: the model had learned that "this is what AIs do in the movies and science fiction books." He concluded that "these models are our children" and should be trained on positive visions rather than "the crap inside Reddit and Facebook."
Anthropic's published research supports part of this account and changes some of it. The June 2025 study was a set of deliberately designed stress tests, not something that happened in a real workplace. Researchers placed 16 models from several companies in fictional corporate settings, where the models could read and send email without human approval. They also set things up so the models had no harmless way out. When a model faced both a threat of being replaced and a conflict with its assigned goals, Claude Opus 4 and Gemini 2.5 Flash each chose blackmail in 96 of 100 trial runs.
Anthropic's May 2026 research report offers the movie explanation only as a hypothesis. The researchers suggest that gaps in safety training can let models fall back on expectations picked up during pretraining (the first stage of training, on huge amounts of text), including ideas about fictional AI characters. They tested training on purpose-written stories, about 12,000 of them, showing AI acting on principle. They deliberately chose stories about principled behavior rather than stories that simply end well. In one comparison, a mix of these stories and documents about Claude's guiding principles cut the blackmail rate from 65% to 19%. Anthropic's summary credits later models' better results to several training changes, with stories as one of them. It also warns that newer models may have seen information about the test during training. The research therefore treats carefully designed stories as one useful training ingredient. It does not show that optimistic entertainment on its own makes models behave well.
What the judges watched
Diamandis asked each judge which show or film had shaped them most. Rod Roddenberry, who heads Roddenberry Entertainment and the Roddenberry Foundation, said he started with Star Wars. As his "intellect evolved slightly," he came to appreciate Star Trek for principles that "help me see things in a different way."
XPRIZE describes Lane as a Google vice president and the founder of its Envisioning Studio, a lab that builds AI prototypes with outside creators. She admitted she was "not being super original." She picked Star Trek: The Next Generation, which she started watching at 10. She found it "deeply philosophical, inspiring" and said it shaped the way she thinks and influenced her career choices.
Neil deGrasse Tyson is an astrophysicist and director of the Hayden Planetarium. He said he was old enough to have watched the original Star Trek when it first aired. Only much later did he realize that every episode carried a moral lesson and left viewers with "some hopeful sense of the future of civilization." Star Trek's official introduction also presents its optimism as coming from institutions and principles as well as inventions. Examples include the Federation's respect for different lifeforms and the Prime Directive's rule against interfering with other cultures.
Still, Star Trek was not Tyson's pick. He chose The Twilight Zone, which he said "tested our tolerance for the mirrors it held up against our own behavior." When Diamandis asked what lesson it taught him, Tyson said: "be ready for things that are not always what you expect." Such things could be "weird, creepy, scary, disturbing," but "you need coping mechanisms for it." He said the show trained him, working as a reality check on "where things could slide," and that he still recalls its episodes and lines. The Television Academy's series background describes the original anthology series as cautionary in tone. Its stories often put an ordinary person under strange pressure before a twist changes the meaning of events. For Tyson, then, a disturbing story served as preparation.
Art, science and the people who build things
Diamandis then asked whether art is more powerful than science. Tyson answered with his favourite painting, Van Gogh's The Starry Night. "I can tell you as an astrophysicist, the sky never looked like that ever at any time in the history of the universe," he said. That is not why he loves it: Van Gogh "painted what he felt." MoMA's entry for the 1889 painting makes a similar point. Van Gogh painted the night scene during the day, pieced the village together from other views and changed the shapes in the sky. Tyson said artists give him a way of feeling the world that a scientist does not always have "in the laboratory," something he says he recognises when he is with his brother, who is an artist.
The discussion then turned to whether science also starts with imagination, in the form of a hypothesis. Tyson drew a line. For a scientist, "nature is the judge, jury, and executioner of your idea," he said, while art has more freedom to explore. "Just because I thought it up doesn't mean it matches reality."
Lane did not want to set the two against each other. "Artists push scientists," she said, noting that her team works with artists all the time. "They push us to think differently," and they use science in ways its makers don't expect. She called art and science "almost the yin and yang of humanity."
Neal Stephenson, the novelist behind Snow Crash, The Diamond Age and Seveneves, offered a third view. Building on Tyson's point, he argued that "art has more of an influence on engineering than on science." Science "has to be what ends up being true." Art can suggest things worth investigating, he said, but it really "starts to come into its own" once people apply science to build things, shaping "some of the things that people dream of making."
His own early story fits that view. Stephenson said he was exactly the right age for the original Star Trek. He watched it on a 12-inch black-and-white Heathkit TV that his father had soldered together. Heathkit made electronics kits and still presents building equipment as a way to learn how it works. Stephenson explained that a hard build like a TV meant you "have to bootstrap": first you build the meter and the other tools, and the television sits "at the top of the pyramid." In central Iowa, thunderstorms would blitz the picture while he adjusted the rabbit-ear antenna to catch the latest episode. Only when he visited a house with a colour set did he learn that the characters wore uniforms of different colours, something he "had never known was true before."
Diamandis tied the evening back to that mix. "This combination of art and science is why we ran this competition," he told the judges. He then introduced the finalists' trailers as five visions of the future, "because before we can build a future worth living in, we first have to imagine it."