Training AI requires more data than we have — generating synthetic data could help solve this challenge

By A.T. Kingsmith, Lecturer, Liberal Arts and Sciences, OCAD University

The rapid rise of generative artificial intelligence like OpenAI’s GPT-4 has brought remarkable advancements, but it also presents significant risks.

One of the most pressing issues is model collapse, a phenomenon where AI models trained on largely AI-generated content tend to degrade over time. This degradation occurs as AI models lose information about their true underlying data distribution, resulting in increasingly similar and less diverse outputs full of biases and errors.

As the internet becomes flooded…

Read complete article

~	What is aflatoxin, the toxic chemical behind Coles’ peanut butter recall?

~	Parents are feeling anxious about men in childcare centres. Stronger rules would make everyone feel safer

~	Australians will soon need their age checked to log into online search tools – here’s why

~	Trump’s ‘big beautiful bill’ has passed the US Senate – these are the winners and losers

~	Silenced, transferred, threatened: Women are still speaking out in post-uprising Bangladesh

~	K'iche’ Maya glyphs to reconnect art and language: Wilmer Aram Ajú's project in Guatemala

~	Distrust in AI is on the rise – but along with healthy scepticism comes the risk of harm

~	More than meds: why easier access to ADHD treatment has to be part of a whole-system approach

~	What’s next for NSW’s intense storm? Heavy rains, fallen trees – and a chance of a storm ‘slingshot’

~	What are working with children checks? Why aren’t they keeping kids safe at daycare?