Bots vs Docs: Who's Winning the Bedside Manner Battle?

Today:

AI chatbot's 'bedside manner' preferred over conventional doctors by shocking margin

Well, butter my biscuit! Nearly 80% of folks are now rooting for a machine over a good ol' human doc, says a blind study from The Journal of American Medicine. Seems people are mighty pleased with the 'bedside manner' of this AI chatbot, ChatGPT, over their flesh-and-blood doctors.

Chris Winfield, the bigwig of Understanding A.I., shared these tidbits on "Fox & Friends Weekend". He reckons folks might be fed up with doctors being as overworked as a one-legged man at a butt-kickin' contest. No surprise there, what with the way healthcare's going these days.

With AI storming the field like a bull in a china shop, it's helping doctors spot early signs of some real nasty stuff like Alzheimer's, strokes, and cancer. But it also has folks scratching their heads over whether these machines might just take over the whole dang profession.

But hold your horses, not everything's as peachy as grandma's pie. There's this worry that AI could be as biased as the humans feeding it data. Not to mention, there's the looming question of whether AI could leave folks without a job. Just goes to show, even the shiniest apple can have a worm in it.

So, folks are left wondering: Is AI the best thing since sliced bread or just another horse of a different color? Time will tell, I reckon.

READ THE ARTICLE ON FOX NEWS.

The Amazing Ways Duolingo Is Using AI And GPT-4

Alright, buckle up folks, we're about to dive into the world of Duolingo, the app-based learning whiz kids that are pushing the envelope in language learning. Now, they're not exactly new kids on the block, they've been at it since 2011 with the grand aim of being your personal language tutor. But here's where it gets juicy: they've recently teamed up with OpenAI to use GPT-4, the latest smarty-pants AI model, to teach us all how to order croissants in French.

If you've been living under a rock, GPT-4 is the latest hotshot AI model from OpenAI that's been making waves. It's been trained on one trillion parameters, which is a fancy way of saying it's learned from a LOT of data. Now, Duolingo is one of the first to put this new tech to use.

So, what does this mean for us, the humble language learners? Well, Duolingo has crafted a couple of nifty features. The first is a role-playing game where you can chat with an AI barista in Paris, maybe practice your coffee ordering skills, or just shoot the breeze. The second feature's called "Explain My Answer" and it's like having your own private tutor. Got a question wrong? No worries, this feature will break down exactly where you tripped up.

Sounds easy-peasy, right? Not so fast. Building these features was no walk in the park. They had to work closely with OpenAI, generate and label a ton of data, and refine the prompts being sent to the AI model. They also had to put the kibosh on any unexpected or undesirable results, making sure the AI stays on track and provides a safe, engaging learning experience.

What's next for Duolingo? They're not stopping at croissants and coffee. They're exploring how AI can generate new course content and become even better at tailoring the learning experience to individual users. All in all, they're aiming to make language learning a piece of cake for everyone, and they're using AI to do it.

So, there you have it. Duolingo, OpenAI, and GPT-4, making the world a smaller place, one language lesson at a time. I tell ya, it's enough to make your high school French teacher green with envy.

READ THE ARTICLE ON FORBES.

Prepare for Game-Changing Google AI Products After Google Brain and DeepMind Merge to Compete With OpenAI

Google's playing mix-and-match with its AI brainboxes, mushing together Google Brain and DeepMind to form a new dream team: Google DeepMind. It's a move to keep up in the big leagues of AI, with OpenAI's ChatGPT batting in the opposite corner.

This shiny new AI crew is taking the wheel of Google's future tech innovations. The boss man? Demis Hassabis. He's stoked about pooling top-drawer AI talent and Google's hefty resources to cook up some next-gen AI magic.

Hassabis ain't pulling this out of thin air, though. Google Brain and DeepMind have already done some heavy lifting in the AI game, from deep reinforcement learning to transformers. And now, with the new merger, they're looking to whip up the next big wave of AI goodies.

In an office-wide memo, Hassabis basically said, "Buckle up, folks. We're going to change the world (and maybe even outpace OpenAI)."

This ain't DeepMind's first rodeo. They're the folks behind AlphaGo, who made pro human Go players cry, and AlphaFold, that crazy accurate protein predictor. They've also got Jeff Dean, Google's head honcho scientist, joining the squad.

After Google shelled out a cool $500 million for DeepMind in 2014, it wasn't all sunshine and rainbows. There were a few tiffs, especially when Google started nudging DeepMind to turn its research into cash cows.

Google DeepMind is the tech giant's pledge to keep pushing the envelope in AI research. And with the heat on in the AI market, Google's betting on DeepMind to bring home the bacon.

Google ain't sitting on its hands, either. It's been hammering away at its own AI called Bard to go toe-to-toe with Microsoft's GPT-powered chatbot. But so far, Bard's been more of a dud than a stud, even causing a $100 billion hiccup for Google when it botched some info in a promo video.

But don't count Google out just yet. They're known for their hush-hush product development. So, who knows? Google DeepMind might be cooking up something big enough to shake up the whole AI scene.

READ THE ARTICLE ON BUSINESS2COMMUNITY.

Arize launches Phoenix, an open-source library to monitor LLM hallucinations

Arize AI, the California whiz-kids behind machine learning observability, just unleashed Phoenix, a fresh-off-the-press open-source library made to keep an eye on those big-brained language models when they start seeing things. Sounds like an AI babysitter, right?

You see, these language models are the new shiny toys for data scientists. They're being used in everything from digital lawyers to healthcare chatbots. But as cool as they are, they're prone to "hallucinations" - basically, they can spit out some seriously wrong info. That's where Phoenix swoops in.

The big idea? Phoenix lets you watch these language models think, flagging when they trip up or start talking nonsense. Imagine it like a football coach watching game tapes, picking out every fumble and wrong turn.

Jason Lopatecki, the big cheese at Arize, breaks it down like this: you load up your chatbot conversation into Phoenix, and the software maps out the chat, scores each response, and shows you where your bot was a superstar and where it fell flat on its face.

This nifty visual can then be used to dig deeper, find problem areas (like when your bot starts speaking gibberish to Spanish speakers), and fine-tune the model for better results. It's like giving your bot a tune-up based on its driving history.

Phoenix is stepping onto the scene as the first of its kind, helping users understand and manage these complicated language models. But this is Silicon Valley we're talking about. New players are popping up like weeds, so keep your eyes peeled.

Even though AI models might seem like a locked mystery box to most folks, Phoenix is designed to give you a peek inside, mapping out how these models think and decide. It's been given the thumbs up by 100+ users and researchers, but only time will tell if it really lives up to the hype.

Christopher Brown, another tech bigwig, called Phoenix a "much-appreciated" step forward. He believes it'll save time and push teams to think harder about improving their models before sending them out into the wild. So, here's hoping Phoenix does more rising than burning, eh?

READ THE ARTICLE ON VENTUREBEAT.

AI race spurs new era in web browser wars

Well, folks, it's a wild, wild west in the world of web browsers again, and this time, it's the AI sheriff that's calling the shots. All the tech giants are throwing their hat in the ring, betting big on AI to spice up your web surfing experience.

Microsoft, the old warhorse, is sprucing up Bing and Edge with a new AI tool called ChatGPT, developed with OpenAI. It's a power move that's got Google's pants in a bunch, reportedly prompting them to work on AI updates for Chrome under the codename "Magi". Even Mozilla's joining the fray, launching a new AI-focused startup called Mozilla.ai for Firefox.

But here's the skinny: despite the browser buffet on offer, Google Chrome's still the big cheese, owning almost 65% of the market. Safari's playing second fiddle at 19.5%, with Edge and Firefox bringing up the rear at 4.6% and 2.9% respectively.

That said, Microsoft's been chalking up some gains. Edge has been stealing market share for eight straight quarters, and Bing's been growing in the US. Even got Samsung considering switching their default from Google to Bing. High drama, indeed!

Google and Mozilla are not sitting on their hands, though. Google's doubling down on AI, merging their Brain team and DeepMind to turbocharge their AI game. Mozilla, on the other hand, is playing the "trustworthy and responsible AI" card, promising an alternative that prioritizes human well-being and transparency over profits.

While these tech giants duke it out, we're stuck in the peanut gallery waiting to see who'll come out on top. Here's hoping this AI showdown brings us some real-deal improvements and not just a bunch of bells and whistles.

READ THE ARTICLE ON FOX BUSINESS.

G7 should adopt 'risk-based' AI regulation

Alright, let's chew the fat here. The bigwigs from the G7 got together in Japan, and they're all nodding heads about this "risk-based" AI regulation thing. Now, they're not saying "halt all progress," but they do want to keep an eye on any potential monkey business.

These digital honchos are keen on making sure new tools, like that chatbot ChatGPT, play by the rules. They're playing it cool though, they still want an "open and enabling environment" for AI. You know, freedom, democracy, all that jazz.

They're saying "we're all in this together," but they're also saying each G7 country might handle things a bit differently. Gotta respect the home turf, after all.

ChatGPT, the talk of the town and fastest-growing app since its debut, has them especially worried. They're planning more chinwags about it, touching on topics from copyright to disinformation.

Italy, one of the G7, pulled a fast one and shut down ChatGPT last month, questioning if it stepped over personal data lines. The ban's lifted, but it's got other European countries sniffing around now.

EU's tech regulation head honcho, Vestager, says they're working on this AI Act, including protecting copyright for generative AI. Also, labeling obligations for AI-generated stuff. Gotta keep the teachers and artists safe, right?

Japan, hosting the G7, is playing the good cop, cheering on AI developers and pushing for flexible governance instead of a "one-size-fits-all" regulation. Meanwhile, France is being a pal to the little guys, promising exceptions for small AI developers.

On a more serious note, they've got their knickers in a twist about security risks. Our Japanese friend Kono pointed out that generative AI can stir the pot if it's fed dodgy data. Fake news, anyone?

All these tech titans - from Britain, Canada, the EU, France, Germany, Italy, Japan, and the US - convened in Takasaki, a stone's throw away from Tokyo. The G7 Summit's coming up in Hiroshima in late May, where they'll continue to hash out AI rules. Stay tuned, folks!

READ THE ARTICLE ON REUTERS.

Lawmakers introduce bill to block AI from launching nuclear weapons, push for more human control

So, here's the skinny: Our lawmakers have gotten together to say a big, fat "Nope" to AI having any say in launching nukes. They've cooked up a bill called the Block Nuclear Launch by Autonomous AI Act, because they reckon the current ban by the Department of Defense just doesn't cut the mustard.

The gang of four behind this bill are Senator Edward Markey, and Representatives Ted Lieu, Don Beyer, and Ken Buck. They want to make sure no greenbacks are used to let some robot arm press the big red button without a living, breathing human giving the nod first. Because, let's face it, letting machines decide when to go all kaboom isn't the best idea we've heard.

The bill also makes it clear that no autonomous system should be picking targets for a nuclear strike. They're saying, "Listen here, AI, any decision to launch a nuke should be made by humans, capisce?"

And while using AI for launching nuclear fireworks is already a no-go, these folks reckon it's important to make some noise about it, so China and Russia get the memo too. It's also a good chance for them to show they're all in on stopping the spread of nukes.

So, if this bill makes it through, it'll be another win for Team Human. Robots, you're on the bench for this one.

READ THE ARTICLE ON BUSINESS TODAY.

4 tech giants mentioned AI a total of 168 times on earnings calls, showing just how much attention the tech is attracting

All the tech bigwigs are aflutter about artificial intelligence (AI), and they're not keeping it under wraps. The four horsemen of tech, Meta, Alphabet, Microsoft, and Amazon, couldn't stop yapping about AI on their recent earnings calls – 168 times, to be exact. Guess it's the new black in Silicon Valley!

Mark Zuckerberg, Sundar Pichai, Satya Nadella, and Andy Jassy, like proud papas, bragged about how they're using AI to spruce up products and rake in the big bucks.

Alphabet took home the blue ribbon, blabbing about AI 64 times. Pichai was all about embedding "deep computer science and AI" in product updates, despite some backlash about their hastily launched chatbot, Bard.

Microsoft wasn't far behind, dropping the AI bomb 50 times. They're so gung-ho, they're betting $10 billion on AI through their stake in OpenAI.

Meta, with Zuck taking up the AI cheerleading squad, mentioned it 47 times. They're using AI to whip up some fancy "visual creation tools" for Instagram. Expensive? You bet. But Zuck's not sweating it.

The e-commerce behemoth, Amazon, was a bit more coy, only saying AI seven times. But don't be fooled, they're in for the long haul, ready to spend "billions" and "many years" developing language models for chatbots, according to Jassy.

So, folks, it's clear: AI's the belle of the ball for these tech tycoons. And they're ready to spend big to stay ahead in the game.

READ THE ARTICLE ON BUSINESS INSIDER.