ChatGPT-4 Breaks New Ground: 40% Success Rate in Tricky Medical Diagnoses

Use ChatGPT to Comment with 1 Click on LinkedIn Prospects

Join 15,000+ LinkedIn power users and supercharge your social selling with Engage AI! Imagine effortlessly writing insightful comments that break the ice and build relationships with prospects. With Engage AI as your comment writing assistant, you'll save precious time while achieving conversions with every added touchpoint. Give Engage AI a spin and transform your LinkedIn conversations into powerful conversion tools today.

Today:

ChatGPT AI Shines in Challenging Medical Cases

A team of smart people at Beth Israel Deaconess Medical Center ran some tests on GPT-4, a chatbot, to see if it could diagnose tricky medical cases. GPT-4 hit a home run about 40% of the time with the right diagnosis straight off the bat, and for nearly two-thirds of cases, it was somewhere in the ballpark with the correct diagnosis in its list of possibles.

They tested GPT-4 using a bunch of tricky patient cases from the New England Journal of Medicine, typically used to make doctors scratch their heads. GPT-4 was spot-on in 39% of the cases and had the right answer in its list of possibilities in 64% of the cases.

So, while this isn't saying we can replace doctors with a chatbot, it's an interesting peek at how AI might lend a hand to doctors, helping them make sense of complex data and broaden their thinking on diagnoses.

READ THE ARTICLE ON NEUROSCIENCE NEWS

Google StyleDrop generates images from text

Google's brainchild, StyleDrop, can whip up a visual representation of the gibberish you throw at it. Just pen down what you want to see and the style you're craving for, and voila, you'll be served with your custom art, piping hot in about three minutes. It's like having your own creative genius, but without the messy studio.

Just say you want a bridge, and then decide on the dressing: "make it melt like Dali's clocks" or "give it that 3D pop". The creative liberty is all yours. Heck, even your typography can play dress-up. Want your letters to mimic the style of your melting masterpiece? Done and dusted.

Google's other brainchild, Muse, buddies up with StyleDrop for this artistic endeavor. Muse knows a thing or two about photorealism, thanks to its training with 3 billion parameters. Now, you might think, "that's a whole lotta parameters." And you'd be right, but that's what gives Muse its remarkable knack for creating high-quality images.

READ THE ARTICLE ON TECHXPLORE

New Tool Reveals How AI Makes Decisions

German brainiacs at Darmstadt and Aleph Alpha have crafted a tool called AtMan that's breaking open the secret language of artificial intelligence (AI). AtMan is a digital bloodhound, sniffing out why AIs make certain decisions. It’s a chatty Sherlock Holmes, laying bare the evidence but still not revealing how the crime was planned.

AtMan also doubles as an AI proofreader and a bouncer, helping to review AI-generated summaries and spot bias. It's a handy little fellow even for image-related AI tasks - kinda like a helpful AI pointing out abnormalities in a medical scan. But make no mistake, the world of AI is a web of mechanical brain copies, as complex as a high-stakes game of telephone.

READ THE ARTICLE ON SCIENTIFIC AMERICAN

Mechanical Turk workers are using AI to automate being human

Almost half the workers on Mechanical Turk - a platform designed to outsource simple tasks to humans - are actually using AI to complete their assignments. Yes, folks, we've finally arrived at the AI doing the work meant for humans on an AI platform.

Mechanical Turk, an Amazon service, was intended as a means for users to distribute minor tasks that take a few seconds each. Dedicated workers could rack up the pennies and make a fair buck by doing loads of these tasks. All this, in Jeff Bezos' eloquent phrasing, was called "artificial artificial intelligence."

This whole mess unraveled when the researchers contemplated using Mechanical Turk for improving or fact-checking AI responses. But if the very humans fact-checking AI are using AI themselves, where does that leave us? This question led the researchers to estimate that a whopping 33%-46% of workers were employing AI for a task.

READ THE ARTICLE ON TECHCRUNCH.

UNIDO Announces AI Alliance with Huawei and global partners to help Industries Go Digital

UNIDO, the United Nations Industrial Development Organization, just gave the green light for an AI alliance with Huawei and other global heavyweights. The gist? They're all itching to give industries a digital facelift.

This announcement happened in Hungary at the 2023 Intelligent Rail & Logistics Forum. They want to build a playground for sharing ideas, cooking up new ways to use AI in industry and manufacturing, and give the digital economy a shot of adrenaline.

Ciyong Zou took the mic and gave a rousing speech. It was all about the benefits of 5G, AI, and cloud computing coming together like PB&J. He talked up the improvements these tech advancements have brought to transportation and logistics. More efficiency, better job quality, safer work environments, and more women working in sectors that were once boys' clubs.

READ THE ARTICLE ON HUAWEI

Elon Musk ‘didn’t think anyone would actually agree’ to the A.I. pause he called for

Back in March, Musk and a merry band of tech titans signed an open letter requesting a six-month hiatus on A.I. fancier than OpenAI's GPT-4 chatbot, citing doomsday-level risks. To his surprise, people sat up and took notice.

However, not everyone was on board with the plan. Bill Gates, Senator Mike Rounds, and A.I. heavyweight Geoffrey Hinton thought the suggestion was as useful as a chocolate teapot. The logic? If the U.S. hits the brakes, China will keep the A.I. show rolling. As Google CEO Sundar Pichai said, it's a group effort issue.

Aidan Gomez, the head of a $2 billion A.I. startup, Cohere, voiced a similar sentiment, calling the call as achievable as herding cats. According to him, fretting over an A.I. uprising is like worrying about a cow jumping over the moon – a hilarious misuse of brainpower.

READ THE ARTICLE ON FORTUNE