AI Generated Entertainment: Sora & Suno

Earlier this year OpenAI introduced the world to Sora. A text-to-video model that can generate realistic videos from descriptive prompts.

The high fidelity and rich quality of the videos generated by Sora captured the collective imagination of the Internet. Was this the first step toward replacing Hollywood? Are we heading to a world where the movies we watch are generated for us in real-time by an AI model like Sora?

I’ve been a skeptic about AI generated entertainment content. I didn’t think AI could capture the “creativity” of human beings. I couldn’t imagine a world where an AI model could produce something like the Dune 2 film.

And yet, I’m now beginning to believe.

Suno changed my mind.

Suno.ai is a generative music creation program. No instrument needed. Type in your prompt and receive a fully produced professional sounding music track.

Here are two examples:

Fall of a Hero https://app.suno.ai/song/f18445a8-ed57-41bc-8b6a-18e33fc85560

Prompt: Movie soundtrack. Symphonic orchestra, Chord Progression, Main theme, Viola solo, Female ethereal humming, Emotional

Mozart On The Bass https://app.suno.ai/song/063f3392-fe1b-4ce9-9ac8-660dd0315bfb

Prompt: 1750s dubstep, chanting, heavy male vocals, hardstyle, metal fire rap Kpop crazy

The quality of these are stunning.

Add these tracks to any Spotify playlist and I’d wager that listeners wouldn’t realize they were listening to an AI generated song. They may even “favorite” the tracks, look up the artists and “follow” them. And that insight, that the songs being AI generated in 2024 are indistinguishable from human created songs, is what has made me a believer in AI generated entertainment.

Audio seems to be the perfect medium to get us started.

Relatively speaking, it’s “easier” to create a 3-minute pop song compared to a 120 minute Hollywood blockbuster. And given the quality of songs being produced by Suno today, is it feasible to believe that within the next 5 years an AI generated song is nominated for a Grammy?

Earlier in this post I wrote that before Suno, I couldn’t imagine a world where an AI model could output a film like Dune 2.

But is it such a far cry to imagine that 10 years from today, I’ll be able to prompt Sora with the following:

Using the text in the book “Dune” by Frank Herbert, generate a 4 hour epic R-rated movie in the style of James Cameron with Ryan Gosling as Paul and Emma Stone as Jessica.

I used to believe this wouldn’t happen. But now I can’t see it not happening.