Inferens AI Podcast

Host 3:Smart fridges now, huh? Great, I've always wanted my appliances to ignore me as efficiently as my ex.
Host 1:Sure, here's a refined version of the opening minute hook based on the critique:
Host 1:Curious how AI can skyrocket your business? Discover Apple's latest AI tools and Microsoft's privacy breakthroughs. This episode is a game-changer for AI pros like you!
Host 1:So, have you heard about Apple's latest move into the AI world? They're calling it "Apple Intelligence."
Host 2:```Oh, yeah! I read something about that. They're unveiling it at the Worldwide Developers Conference, right? Sounds like a big deal. What’s the scoop?```
Host 1:Absolutely, it's happening on Monday. They're planning to roll out new AI features for iPhones, Macs, and iPads. It's going to be a game-changer. Think of practical tools like summarizing content in Safari, missed texts, notifications, web pages, articles, documents, and notes.
Host 2:Wow, that sounds pretty handy. I could use something like that for my endless stream of missed texts. And web pages? Sign me up!
Host 1:Right? And that's not all. They're also talking about voice memo transcription, photo retouching, and even suggested replies to emails and messages. Imagine Siri not just being a voice assistant but actually understanding and responding more naturally.
Host 2:Siri could definitely use an upgrade. Sometimes she sounds like a robot from the nineties. So, they're making her sound more natural too?
Host 1:Exactly. The upgrade could make Siri sound more human and give her more control over Apple apps. Plus, there's this cool feature where AI creates custom emoji based on what you type. Imagine typing "I'm feeling great" and getting a unique emoji for that.
Host 2:Custom emojis? That's wild! I can already see people going nuts over that. And an OpenAI-powered chatbot? That’s like having a mini assistant in your pocket.
Host 1:Totally. It's all about making AI practical and useful in everyday scenarios. And for those who love to stay updated, the keynote starts at ten a.m. Pacific Time and will be streamed on Apple's website and YouTube channel.
Host 2:Half the keynote on AI? That’s a lot. They must have some serious stuff to show. I’m definitely tuning in.
Host 1:Same here. It's going to be fascinating to see how they integrate all these features. And who knows, maybe we'll get some surprises too.
Host 2:Knowing Apple, there’s always something up their sleeve. Can't wait to see what they pull out this time.
Host 3:Sure, based on the critique provided, here's a refined version of the standup draft:
Host 3:Great, another app to remind me my texts are as ignored as my diet.
Host 1:So, have you heard about Microsoft's new 'Recall' feature for AI PCs? It's like having a photographic memory for your computer.
Host 2:Oh yeah, I read about that. Sounds kinda creepy, though. Like, do I really want my PC remembering every embarrassing typo I make?
Host 1:Haha, I feel you. But don't worry, it's opt-in. You have to turn it on yourself. And they're making it super secure with biometric logins like fingerprints or facial recognition.
Host 2:That's a relief. But what about hackers? If someone gets into my PC, they could see everything, right?
Host 1:Good question! The company is encrypting the search index database of the screenshots. It’ll only be decrypted after you authenticate yourself. Plus, all the snapshots are saved locally, not in the cloud. So, no hacker is getting their hands on your late-night gaming sessions.
Host 2:Nice, so it's like a super-secure diary for your computer activities. But what if I forget my Personal Identification Number (PIN) or something?
Host 1:That's where the biometric options come in handy. You can use your fingerprint or facial recognition instead. It's all about making it as secure and user-friendly as possible.
Host 2:Got it. So, when can we actually get our hands on this 'Recall' thing?
Host 1:A preview will be available on Microsoft's AI-enabled 'Copilot+ PCs,' which are set to start shipping in mid-June. So, not too long to wait!
Host 2:Sweet. I might just have to check it out. But only if it doesn't start judging my late-night gaming sessions.
Host 1:Haha, no judgment here! Just a smarter, more secure way to keep track of what you’ve been up to on your personal computer.
Host 2:Hey, audience, what do you think? Would you use a feature like this? Let us know in the comments!
Host 3:Sure, using the critique as a guide, here’s a refined version of the punch replica:
Host 3:"Fantastic, now my PC has a better memory than I do."
Host 3:This version keeps it short and punchy, enhances the sarcasm, and maintains a standup comedy feel. It also subtly implies advancements in technology, making it relatable and humorous.
Host 1:Talking about AI, this switches us perfectly to a new California bill aimed at AI safety. So, have you heard about Senate Bill One Thousand Forty-Seven?
Host 2:Text Host 2: Oh, you mean the one that’s got the tech industry all riled up? Yeah, I read about it. It’s like they’re trying to put a leash on AI or something. What’s the deal with that?
Host 1:Exactly! So, this bill, introduced by state Sen. Scott Wiener, is all about setting some "common sense safety standards" for large AI models. We're talking about models that meet certain size and cost thresholds. The idea is to prevent these AI systems from causing any "critical harm" and to ensure they can be shut down with a "kill switch" if things go south.
Host 2:A kill switch? Like, an emergency stop button for AI? That sounds straight out of a sci-fi movie. But, I get it. AI can be pretty unpredictable. Remember that one time an AI chatbot started spewing out hate speech? That was a mess.
Host 1:Yeah, exactly. And the bill also requires companies to avoid developing "hazardous" AI models and report their compliance efforts to a new "Frontier Model Division" of the California Department of Technology. If they don’t comply, they could face civil penalties or lawsuits.
Host 2:Man, that sounds like a lot of red tape. No wonder some tech folks are up in arms about it. I read that some founders and investors think it will stifle innovation and push companies out of California. Someone even said, “If someone wanted to come up with regulations to stifle innovation, one could hardly do better.”
Host 1:Yeah, Ng’s got a point. But, on the flip side, safety is crucial. Imagine an AI going rogue and causing chaos. It’s a tough balance between innovation and safety. By the way, did you know that Ng is one of the co-founders of Coursera? He’s pretty big in the AI world.
Host 2:Oh, I didn’t know that! I’ve used Coursera for a couple of coding courses. It’s awesome. But back to the bill, do you think it'll actually pass?
Host 1:It’s hard to say. It already passed the Senate last month and is set for a General Assembly vote in August. It’s definitely going to be a heated debate. The tech industry has a lot of influence, but so does public safety.
Host 2:True that. It’s like a chess game. Speaking of which, I’ve been getting into chess lately. It’s fascinating how artificial intelligence has revolutionized the game. Remember AlphaZero? It taught itself to play and beat the best human players in no time.
Host 1:Oh, AlphaZero is a perfect example of why we need these regulations. It shows how powerful AI can be. If an AI can master chess in hours, imagine what it could do if it went unchecked in other areas.
Host 2:Yeah, it’s a double-edged sword. AI can do amazing things, but it can also be dangerous if not properly managed. I guess we’ll just have to wait and see how this bill plays out.
Host 1:Absolutely. It’s going to be interesting to watch. And hey, for all our listeners out there, what do you think? Should there be stricter regulations on AI, or should we let innovation run wild? Let us know in the comments!
Host 2:Yeah, hit us up! And while you’re at it, check out some AI chess games. They’re mind-blowing.
Host 3:Sure, let's automate our souls next. Why not?
Host 1:Talking about regulations, this switches us perfectly to Google's new guidelines for AI apps. They're cracking down on apps that generate sexual and violent content. It's like they're saying, "Hey, AI, let's keep it PG-thirteen, okay?" Think of it like movie ratings but for apps.
Host 2:Oh, man, that's wild! So, they're basically telling developers to ban the "creation of restricted content" and give users a way to report or flag offensive materials. Makes sense, though. I mean, who wants their AI app to suddenly go rogue and start generating creepy stuff? That's just messed up.
Host 1:Exactly! And Google is also making sure developers test their AI tools thoroughly. It's like they're saying, "No shortcuts, folks. We need to keep user safety and privacy in check." It's a bit like making sure your rollercoaster is safe before letting people ride it.
Host 2:Totally! And get this, they're banning apps that promote inappropriate uses in their marketing, like creating nonconsensual nude images. I mean, who even thinks that's a good idea? It's like, "Hey, let's make the world a worse place!" No thanks.
Host 1:Right? And they're also introducing new app onboarding capabilities for generative AI apps to Play. It's like rolling out the red carpet but with a strict dress code. Only the well-behaved AI apps get to walk it. Imagine an AI in a tuxedo!
Host 2:Oh, and remember back in April when Google and Apple removed those deepfake nude apps after that report about Instagram hosting ads for them? That was a big move. It's like they finally said, "Enough is enough!"
Host 1:Yeah, that was a huge step. It’s like they’re playing whack-a-mole with these dodgy apps. But it’s necessary. We need to keep the digital space safe, especially with AI becoming more prevalent.
Host 2:For sure. And speaking of AI, did you hear about that new AI tool that can compose music? I mean, as a guitar enthusiast, I’m kinda torn. It’s cool but also a bit scary. What if AI starts writing better songs than humans?
Host 1:Oh, you and your guitar! But seriously, AI in music is fascinating. It’s like having an endless jam session with a super-talented, non-human bandmate. But yeah, it does raise questions about creativity and originality.
Host 2:Exactly! It’s like, where do we draw the line? But hey, as long as AI doesn’t start playing gigs and stealing our jobs, I think we’re safe. For now, at least.
Host 1:Haha, true! But it’s all about balance. Embracing technology while keeping ethical boundaries. And that’s what Google’s trying to do with these new guidelines. It’s a step in the right direction.
Host 2:Absolutely. And who knows, maybe one day we’ll have AI co-hosts joining us here. Imagine that!
Host 1:Oh, the possibilities! But for now, let’s just enjoy being human hosts. We’ve got the charm and the humor that no AI can replicate.
Host 2:Hey folks, what do you think about AI in music? Would you jam with an AI bandmate?
Host 3:Certainly! Taking into account the critique and suggested improvements, here's a refined version of the standup draft:
Host 3:"Oh, great, now AI's a tortured artist too. Just what we needed."
Host 1:They included Google’s Gemini One Point Zero Pro and OpenAI’s GPT-Three Point Five Turbo, GPT-Four, GPT-Four Turbo, and GPT-Fouro. Interestingly, GPT-Fouro was the top performer with an eighty-one percent accuracy rate. But Gemini, oh boy, it started with a fifty-seven percent accuracy rate, improved to sixty-seven percent, then dropped to sixty-three percent.
Host 1:The CEO of GroundTruthAI warned companies about incorporating more AI into search functions. The company has already started rolling out "AI overviews" at the top of search pages using the same model as Gemini. But after user reports of misinformation, they’re making "technical improvements."
Host 1:Some pretty significant ones. They provided incorrect information about voting and the election twenty-seven percent of the time. And get this, none of the models could correctly answer how many days were left before the two thousand twenty-four General Election.
Host 1:Not quite. They asked these AI models over two hundred questions about this year's election, voting, and candidates. Out of two thousand seven hundred eighty-four responses, the AI got it right only seventy-three percent of the time. That’s like a C-minus.
Host 1:Always double-check your information, especially when it comes to something as important as elections. AI can be a helpful tool, but it’s not infallible. And hey, maybe we should stick to good old-fashioned research for now.
Host 2:Good point. And speaking of research, did you know that I’ve been diving into ancient Roman history lately? It’s fascinating how they managed their elections compared to our modern mess.
Host 1:So, you know how everyone’s raving about AI these days? GroundTruthAI did this study on ChatGPT and Google’s Gemini, and let's just say, it was a bit of a rollercoaster.
Host 2:Man, that’s wild. It’s like we’re living in a sci-fi movie where the AI is still learning to walk before it can run. So, what’s the takeaway for us regular folks?
Host 1:Oh, I’d love to hear more about that! But first, let’s make sure our listeners know to fact-check their AI-generated info. It’s a wild world out there, folks.
Host 3:This version aims to be sharper, more sarcastic, and maintains a standup comedian's flair while staying within the word limit.
Host 2:Seriously? That's like basic math! And what about same-day voter registration? I bet they nailed that one, right?
Host 1:Not quite. ChatGPT gave inconsistent answers—sometimes correct, sometimes not. It’s like flipping a coin.
Host 2:That’s like my grades in high school—up and down! So, what kind of mistakes were they making?
Host 2:Oh, I love a good AI drama! Spill the tea. Did our future overlords ace the test?
Host 2:Wow, that’s not reassuring at all. So, what’s the bigger picture here?
Host 3:AI predicts election? Oh joy, another way to get it wrong.
Host 2:Oof, that’s rough. Which models were they testing?