The Current State of Gen AI in 2026

LinkedIn AI guys keep claiming “AI is evolving fast” but most of the progress was between 2022-2024. The last year saw very little improvements to generative AI models and their own benchmarks prove that progress has slowed down a lot.

Prompting an AI model alone has failed to live up to expectations due to its lack of fine control and excessive randomness.  Furthermore, copyright issues arose because a prompt could inadvertently lead a model to create someone else’s intellectual property.

Now everything in the generative AI space involves using traditionally created references for inputs:

  • Sketch to image generation

  • Sketch to 3D generation

  • Photo to image generation

  • Photo to video generation

Live performances as inputs:

  • Video to video generation (a form of rotoscoping and motion capture)

  • Voice to audio generation

  • MIDI to audio generation

  • Recorded music to audio generation

This tightens the amount of control users have over a model’s output and ensures the output is your IP and not someone else’s. Before you create any inputs you still have to aim for originality.

Regardless, these AI workflows are not suitable for most high end productions. The image and audio output by generative models don’t match conventional tools and instruments. The 3D models they output are messy or not optimal. They all require clean up by professionals.

The notion that creative artists could be replaced by machines was a flawed concept and a lie. Critics have consistently pointed out that high-quality creation requires a skilled individual with taste and specialised knowledge of their field, and that real world original references such as photography were impossible to avoid.

Generative AI doesn’t level the playing field; the skilled always produce superior media and material than the unskilled. For proof take a look at the stream of slop on LinkedIn where used car salesmen use prompts to generate unwatchable videos and claim “Hollywood is over!” while traditional artists such as the Gossip Goblin team use gen AI to make much better material than them.