No mere human can keep up with the torrent of intriguing/terrifying daily AI news, but we’ll try anyway — at least until we’re rendered obsolete by the machines…
***
AI can probably crack your password — and faster than you think
For every lovely, life-affirming application of AI, there’s at least one adversarial, uh-oh-inducing one. This one’s the latter.
Cybersecurity firm Home Security Heroes put an AI-powered password-cracking tool called PassGAN to the test against 15m+ passwords.
Before you reset every password you’ve ever had, remember that usage of PassGAN isn’t widespread — and could also be used to generate crack-proof passwords.
Still, Home Security Heroes recommends stronger passwords with 15+ characters that avoid patterns like “1234.”
***
AI can kinda see what your brain sees?
A team at Osaka University used a deep-learning AI model called Stable Diffusion (SD) to analyze brain scans after people were shown images inside an MRI machine.
The AI was asked to “translate” the subjects’ brain activity into a readable format.
Per the Osaka team, this isn’t “mind reading” — it merely means AI can reproduce images a person has viewed. Which ain’t nothing, according to every tightly clenched muscle in our bodies.
Practical applications are minimal today, but SD research (and subsequent ethics debates) will continue.
***
AI brings a creative crisis to the fortune cookie industry
Every year, ~3B fortune cookies are made globally. Most of them house a message penned by a real-life human.
A startup called OpenFortune Inc. may jettison that human touch, using ChatGPT to generate new messages at rates no copywriter could match.
Per The Wall Street Journal, a divide is forming in the fortune biz between accepting AI’s time- and cost-saving ways, and holding true to self-written tradition.
… But who provides wisdom for the wisdom makers?