Advertisement
Science
OpinionComment
Dennis Lee
Dennis Lee

Opinion | Can art generated by an unfeeling AI ever hope to move us?

  • Generative AI can produce art and design much faster and more perfectly than humans ever could
  • But it may be too reliant on biased search results from the internet, and it can never come up with truly great art – because it cannot feel like humans do

Reading Time:4 minutes
Why you can trust SCMP
1
Illustration: Craig Stephens
Dennis Lee
This is a pivotal year for creative professionals: technology experts have declared it a breakout year for generative artificial intelligence.
Advertisement
Computers already outperform humans in solving complex problems, albeit limited to narrow tasks. We had a Sputnik moment when IBM’s Deep Blue defeated then-world chess champion Garry Kasparov in 1997, and again in 2016 when Google DeepMind’s AlphaGo defeated champion Lee Sedol in a Go match.

We were not worried. Because while supercomputers might foresee the next 100 moves in a finite environment with rules and restrictions, they could not generate creative content from scratch.

But the day of reckoning has apparently come, with generative AI that can develop fresh content through self-improving algorithms. Technology is crossing the “last frontier” of intelligence, threatening all creative fields.

In the art and design world, we now have AI-powered visualisation plug-ins such as Veras, floor-plan adjustment software such as Testfit, and text-to-image tools such as Stable Diffusion, Midjourney and DALL-E2. We are advised to retrain to be manipulators of text prompts or left to risk losing our jobs.

Advertisement

Political scientist Ian Bremmer has described AI as a “weapon of mass disruption” and a top global risk this year. Over the years, other bright minds have also warned about the dangers of AI, including physicist Stephen Hawking, Tesla CEO Elon Musk and DeepMind founder Demis Hassabis. Specifically, they warned about the spread of misinformation and biases.

Advertisement