Godfather of AI Says Only Catastrophe May Slow AI’s Breakneck Development
Geoffrey Hinton – known as the ‘Godfather of AI’ — says it may take a major catastrophic event to wake-up the world to the dangers of AI.
Think: An AI-enabled attack on the world’s financial systems, power grid – or a partial nuclear meltdown similar to the Three Mile Island Disaster.
In the meantime, Hinton expects major AI players like Meta, OpenAI and SpaceXAI to continue to race ahead creating AI that might be easily compromised by bad actors.
In other news and analysis on AI:
*New Writer’s Tool From ChatGPT: Animated GIFs: Writers looking to supplement their work with animated GIFs can now get ChatGPT to put those together.
The finished GIF can be downloaded and posted wherever you like – or you can use a link provided by ChatGPT to steer people to your creation.
The new feature is part of an upgrade – dubbed ChatGPT Images 2.6 – being rolled-out to all subscription levels.
*Writer Credits AI Writing for $1 Million in Sales: A South African author says she raked in about $1 million in sales in one year after she started using AI to write her novels.
All told, the AI wrote 370 of her novels during that time, she says.
Observes writer Myles Illidge: “She said the first draft of an AI-generated book isn’t good and must go through several rounds of model-based developmental edits and other fixes.”
*AI Writing A Hit Among Businesses Writing Review Replies: Businesses getting pelted with bad reviews on Yelp, Google Reviews and similar have become common users of AI writing.
Specifically: A new study of 88,000 Google reviews found that an increasing number of businesses are turning to automated writing to whip-up quick replies to both kudos and critiques they find online.
Observes writer Pankaj Aggarwal, founder, Revfer: “Owner replies (to online reviews) have roughly doubled since 2019, and, since 2023, increasingly carry markers consistent with AI writing tools.”
*Novelist Disqualified From Writing Contest for Using AI: Writer Thelyson Orelien has found himself booted from the running for France’s highest literary award.
The charge: Orelien’s novel was shot-through with AI writing, according to the judges.
Too bad. The novel got rave reviews before it was branded an AI work.
*Cognitive Offloading: For Better or Worse, More Students Turning to AI for Thinking: While getting an answer to any question – sometimes correct – is as simple these days as typing it into a chatbot, a new study finds there’s a trade-off to such knowledge acquisition.
Observes the Stanford Report: “The review found that the most detrimental tools were general-purpose chatbots.
“When students used one to study for an exam, it had the potential to decrease their test scores once they no longer had access.”
*The Reason Why ChatGPT Writing is So Conservative?: OpenAI President Greg Brockman says business customers – not consumers – ChatGPT’s biggest customer sector by far.
That could explain ChatGPT’s migration to the conservative, take-no-chances – and some say much less creative – writing it’s churning-out these days.
Essentially, in business, safe – and many say dull — writing too often wins the day, every day.
*The Decline of AI Writing: ChatGPT, Gemini, Claude and Similar Are Simply Churning-Out Lousy Writing Now: Add Ben Hylak, cofounder, Raindrop, to the growing list of AI users who find writing coming from top AI firms completely uninspiring.
Hylak – who also delves deep into AI agents in this piece – blames the people who are training the top AI chatbots these days.
In a nutshell: Their work ends up reinforcing the wrong kind of results, he indicates.
*Hype Factor: AI Shopping Still Not There Yet: Writers looking to offload shopping for smartphones, computers, software and the like to AI agents may want to hold off for a while longer.
A new study of AI shopping by Visa found that on average, AI agents screwed-up simple, “buy this” tasks 43% of the time.
To be fair, the best AI agent completed its tasks successfully 82% of the time.
But when you’re dealing with your own money, you kinda want the computer to get it right every time.
*Hype Factor Squared: AI Agents at Work Deliver 12% Success Rate: A new study finds that AI agents unleashed to perform everyday knowledge work only succeeded 12% of the time.
That’s like hiring a writer who does great on Mondays, but shows up drunk for work the rest of the week — leaving behind a trail of carnage.
Observes writer Manuel Uth: The test doesn’t quiz AI agents on knowledge, but instead puts them in simulated work environments.
Share a Link: Please consider sharing a link to https://RobotWritersAI.com from your blog, social media post, publication or emails. More links leading to RobotWritersAI.com helps everyone interested in AI-generated writing.
–Joe Dysart is editor of RobotWritersAI.com and a tech journalist with 20+ years experience. His work has appeared in 150+ publications, including The New York Times and the Financial Times of London.