The year in AI

The Future. 2022 experienced major breakthroughs in AI development that had been in the works for years. While this new tech is here to make life easier, it’s also raising ethical questions that don’t seem to have any black-and-white answers yet. If society goes with the flow of the current AI wave (rather than push against it), it might find that once the kinks are worked out, this new tech can take human potential to new heights.

Techno optimism vs. pessimism

2022 news feeds were dominated by ChatGPT and text-to-image generators (you already know), as well as a protein structure database and an AI code programmer. Generative AI had the biggest moment due to advances in math and computing power that offered new ways to train the software. Of course, there’s a light and dark side to every AI model, explains The Washington Post.

• ChatGPT, lauded for its human-like responses, has come under fire for propagating misinformation and bias despite OpenAI’s efforts to install safeguards.
• Dall-E 2, Stable Diffusion, and Midjourney create images with greater speed, flexibility, and realism than previous software, but they’ve all been trained on human-made art, leading to debates over copyright and ownership.
• AlphaFold, which grants open access to over 200 million protein structure predictions for scientific research, “could theoretically be put to malign uses by someone looking to engineer biological weapons or toxins,” reports Vox.
• Copilot, which translates basic human instructions into functional computer code, is in the middle of a lawsuit that deems machine learning piracy.

Keep it moving
It’s clear that AI models trained on human-made data are entering murky waters. New laws that protect creators’ work will likely have to go into effect sooner rather than later.

And the invention of any new tech comes with unpredictable effects downstream. If people drive cars even in the face of a possible road accident, they’ll use AI models even in the face of a possible transgression. Only time will tell.