Generative artificial intelligence may be having its banner moment, but the technology existed long before ChatGPT and DALL-E. It began in 2014 with a paper by Ian Goodfellow and several other researchers entitled “Generative Adversarial Networks” (GANs). Goodfellow is a computer scientist who worked for Google (NASDAQ:GOOGL) Brain and Apple (NASDAQ:AAPL) and is currently with DeepMind. Today, his paper has been cited more than 55,000 times and underpins several AI tools.
Nearly a decade ago, Goodfellow uncovered a breakthrough: by using technology to draw on large amounts of data, AI tools can generate “synthetic” data under the right conditions. Over time, with constant training and feedback, the system learns to provide synthetic data closely aligned with the desired output. Today, these synthetic data might include smart contract code, fraud detection algorithms, and of course, hyperrealistic avatars with your face in the metaverse.