Grok’s seven-minute “Odyssey” draws criticism over AI film’s limits

xAI’s Grok Imagine released the first episode of its planned AI-generated “Odyssey” series on Oct. 23, offering just over seven minutes of footage after Elon Musk had promised a feature-length, historically accurate adaptation of Homer before the end of 2026.

The episode was made with London-based Wonder Studios and director Dev Walker. It follows Odysseus as he feigns madness to avoid the Trojan War, then agrees to leave Ithaca after Palamedes and Agamemnon put his son, Telemachus, in the path of his plow. Odysseus joins the fleet, leaving Penelope and Telemachus behind, before the story jumps 10 years to Troy.

The release is an episodic project rather than the full-length film Musk outlined in July. It also does not support the claim of historical accuracy. The story is drawn from the fragmentarily preserved ancient epic Kyprien, rather than Homer’s “Odyssey.” When WIRED asked Grok whether the clip was historically accurate, the chatbot answered no, saying the production was an adaptation rather than a reconstruction of real events.

Grok further said its characters had no historical basis, its modern English dialogue used British accents rather than Homeric Greek or a reconstructed Bronze Age language, and its imagery reflected contemporary cinematic tropes and AI training data more than archaeological evidence. It identified cavalry in a battle sequence as an error against standard reconstructions of the period.

Three creatives made the episode in less than two weeks. Their process involved scripting, planning images for each scene, creating templates for characters, clothing and locations, generating several Grok Imagine video versions, selecting material and editing it into a narrative. The team used detailed prompts to control animal entrances, character positions and cloaks moving in the wind in the plow sequence; editing, sound work and colour grading took about 18 hours.

The production drew criticism for both its dramatic quality and visual continuity. Reviewers cited stiff or forced facial expressions, static bodies, dialogue that did not match facial movement, changes in hairstyles and clothing patterns, shifting object positions and abrupt transitions between scenes. Online viewers also mocked jarring edits, plastic-looking characters and the modern English dialogue, though some called the work progress for AI video.

The Guardian gave the episode one star out of five, describing the imagery as ponderous and inert, with an uncanny-valley quality and graphics resembling a classical-era video game. Its reviewer said the digital voice performances gave characters an even, overemphatic delivery, while dramatic turns lacked motivation and meaning.

TechCrunch similarly found that individual shots could resemble actors sailing on a real sea, but judged the episode to lack artistry, momentum, human interest and emotional reasons to care about its characters. Critics also highlighted the rapid cut from the Ithacans’ departure to their arrival at Troy.

The episode cost several tens of thousands of dollars, a fraction of the roughly $250 million budget reported for Christopher Nolan’s separate “Odyssey” production. The Guardian argued that the Grok release chiefly illustrated the distance between AI video’s low production cost and its artistic quality.

By the Wednesday after the release, the episode had received more than 7.9 million views on X. Wonder co-founder Xavier Collins said the team would address criticism of the boats’ historical accuracy and the first episode’s fast pacing in the next instalment, with the goal that the last episode they make would be “radically better” than the first.