AI Daily News 20250408: 🔍 Google's AI Mode Can Now Answer Questions About Images 👀 Meta Gaming AI Benchmarks? 💥Shopify CEO Says No New Hires Unless AI Can't Do the Job 🫠Google Is Paying AI Talent to Do Nothing 🤖Build an AI Sales Rep with Zapier
18:00 Share

Welcome to another deep dive from AI Unraveled, the show created and produced by Etienne Newman, a senior software engineer and passionate soccer dad from Canada. If you're liking these little journeys into the world of AI, why not hit that like button and subscribe to the podcast on Apple. Today, we're strapping on our scuba gear and plunging into the fast-moving currents of artificial intelligence, specifically as they were on April 8th, 2025.

we've got a whole stack of fascinating developments to unpack, from how the big tech players are positioning themselves to some really cool practical applications and a few head scratchers along the way.

As always, we aim to distill the most important info, giving you a clear picture without that overwhelming info overload feeling. That's the plan. We've sifted through the daily deluge of AI news to bring you a nice, manageable overview. Consider it your cheat sheet to understanding what really matters in this crazy, fast-paced world of AI. Are you ready to jump in? Absolutely. Where should we start?

Well, there's some buzz around Meta and their Lama 4 Maverick model that seems worth exploring first. Some folks are raising eyebrows about how it performed in those benchmark tests. Oh, this sounds juicy. So what's the main issue with Meta's Lama 4 Maverick? Fill me in. The accusation is that the version submitted for the benchmarks wasn't quite the same as the one they released publicly. You know, those benchmark scores are super important, right? They're how we compare different AI models and gauge progress.

So if the benchmark results aren't truly representative of real world performance, well, that's a problem.

I can see why that would be a concern. It's like judging a race car based on its performance on a test track and then finding out it handles totally differently on an actual racetrack. Exactly. And this isn't just some minor detail. If we can't rely on these benchmarks, it throws the whole system into question. How do we know what to trust if even the big players might be, let's say, fudging the numbers? It's like the foundation of AI evaluation is getting shaky. So what's Meta's side of the story? How are they responding to these allegations?

Amad Al-Dali, Meta's VP of GenAI, has come out and said they absolutely did not train the model on any of the test sets used for the benchmarks. They're attributing the performance differences to some bugs that popped up during the rollout, which can lead to inconsistent quality. So they're blaming it on a rushed launch, not any intentional manipulation. Right. But even so, what's the main takeaway here for us as we try to wrap our heads around this situation?

I think this underscores the need for crystal clear standards across the industry. Transparency is absolutely key here. If these kinds of discrepancies can occur, even if they're unintentional, it highlights how important it is to have universally agreed upon methods for benchmarking and releasing these AI models. Agreed.

Let's shift gears now to something that could have a huge impact on how we all work. Shopify CEO Toby Litke has made a pretty bold move with a new hiring policy. Yeah, Litke's basically saying before we greenlight any new position, someone has to prove that AI can't do that job. It's a pretty big statement about the capabilities of AI in the workplace. Wow, that's a very direct approach. It really shows just how much faith they have in AI's ability to revolutionize their business operations. It

It goes beyond just hiring, though. Lutke announced on X, you know, the platform formerly known as Twitter, that reflexive AI usage is now expected of everyone at Shopify. They're even incorporating AI usage into employee performance reviews. So it's not just a recommendation. It's woven into their company culture and evaluation processes. That's a fundamental change in how they view work itself. Think about the bigger picture.

This could signal a massive shift in white-collar work across tons of industries. If companies start prioritizing AI solutions first, it means human roles will have to focus on tasks that require distinctly human skills, things AI can't handle. Job requirements themselves could look radically different in the future. That's a really interesting point. Now let's dive into some of the specific advancements we're seeing in what AI can actually do. Google's been busy making AI-powered search even more impressive. They have.

Google's AI mode is getting a big upgrade. It can now

analyze images and answer questions about them. This is a major step towards a more versatile, multimodal search experience. Wait, ask AI about a picture? Yeah. How does that even work? That sounds unbelievably complex. It's a pretty neat combo of Google Lens, which identifies stuff in an image, and Gemini, their AI model that understands context. So Lens picks out the elements and Gemini figures out how they relate, their characteristics and the overall scene.

upload a photo, and the AI goes beyond just labeling objects. It actually understands them. That's wild. So can I ask you something like what kind of bird is this or what style of architecture is this building? You got it. This isn't fully rolled out to the public yet, though. They're testing it through their labs program to gather feedback and smooth things out before a wider release. Still, it gives a glimpse of the future of search. How do you think this will change the way we find information? It's a huge shift.

We're moving away from typing keywords and towards a much more visual, intuitive, and conversational way of interacting with search engines. Imagine pointing your phone at something and instantly getting all sorts of info in a way that feels completely natural. That's game-changing stuff. Now, let's talk about another area where the competition is cutthroat, attracting top AI talent. Some interesting reports are coming out about how Google's playing this game. Yeah, the news that Google's basically...

paying some deep mind employees to sit on the sidelines is pretty remarkable. It highlights just how fierce the battle for leading researchers is paying people not to work so competitors can't snatch them up. That's a pretty expensive strategy. It is. But it shows just how much these folks are worth in the world. And it's not just that. Google's reportedly using very strict non-compete agreements, especially in the UK, which can stop employees from joining rival companies for a while, even if they're still getting paid.

So how are these researchers reacting to essentially being denched? It can't be great for morale. You're right. There are reports of some researchers feeling sidelined and frustrated. They want to be actively working on AI advancements. Microsoft's VP of AI even mentioned being contacted by some DeepMind folks who feel stuck and are itching to get back to work. This raises some interesting legal questions.

We've seen the U.S. taking steps to restrict non-competes recently, haven't we? Absolutely. The FTC, the Federal Trade Commission, is moving to ban most non-competes in the U.S., but they're still enforceable in the U.K., where DeepMind's main lab is. Google says they only use these agreements in certain cases, though. So the big takeaway here is the incredible value and scarcity of truly elite AI researchers.

Exactly. Companies are pulling out all the stops, even paying people not to work, just to keep this talent in-house. It shows just how much they believe these researchers are key to winning the AI race. Switching gears, it seems OpenAI might be venturing into a whole new area, AI hardware.

The rumors about them possibly buying Joanie Eye's startup are intriguing. Definitely an interesting development. The potential acquisition of IO Products, the AI device company started by Joanie Eye, the guy who designed all those iconic Apple products, and OpenAI CEO Sam Altman suggest OpenAI is thinking big. A rumored price tag of around $500 million tells you they're serious about bringing the design team and their hardware into the fold.

Joni Ive. Wow, that's a huge name to have involved. What's IO Products actually working on? They're focused on building AI-powered personal devices. The details are still under wraps, but think a screenless, smartphone-like gadget specifically designed for seamless interaction with AI. So OpenAI is looking to go beyond software and large language models and get into the hardware game. That puts them more directly in competition with the Apples of the world. For sure.

This partnership between IVE and Altman apparently goes back over a year. They've been actively fundraising and have other former Apple execs on board who know hardware inside and out. The device itself is said to be built by IO Products, designed by IVE's company, LoveFrom, and powered by OpenAI's AI models. This could really shake things up in consumer tech.

OpenAI's cutting edge AI combined with Joni Ives' design prowess, that's a potent combo. What this tells us is that OpenAI might be aiming for a more holistic approach, offering integrated AI solutions that are way more intuitive and seamless than what's out there now. It's a clear sign of their ambition to become a major player in the consumer tech landscape. Now let's go back to Google for a second. They're standing the functionality of their Gemini Live feature, bringing real time visual AI assistance to more users.

Right. This expansion is all about making AI a more useful part of our daily lives. Gemini Live can now process live visual input from your phone screen or camera, giving you help based on what it sees. So I could point my camera at, say, a menu in a foreign language and the AI could translate it and have a conversation with me about it. That's super cool. That's the idea. It's not just about recognizing objects. It's about understanding the context and enabling natural language interaction.

Right now, it's rolling out to some Google One AI premium members and the latest Pixel and Samsung Galaxy phones. Is it as advanced as it sounds? Well, early tests suggest it's more like enhanced Google Lens snapshots, not the continuous video analysis they showed off with Project Astra, the precursor to this feature. Even so, it seems like a big step towards making AI more interactive and practical. Agreed.

Agreed. Real-time visual assistance could be a game changer for everything from DIY projects to navigating unfamiliar places. Shifting gears again, Zapier has released a guide on creating an AI-powered sales representative.

That sounds like it could be a game changer for businesses. Absolutely. Zapier's guide is about building an automated system for managing sales leads with AI. The goal is to streamline the process of capturing, qualifying and nurturing potential customers, all without tons of manual work.

So AI is taking over the tedious parts of sales. Exactly. By combining different software tools with AI, businesses can automate those initial interactions, making sure leads are followed up with quickly and efficiently. This frees up human sales teams to focus on more complex stuff like closing deals. The key benefit being increased efficiency and hopefully higher conversion rates.

That's the core idea. Automating sales can make things much more efficient, lead to more consistent nurturing of potential customers and ultimately a smoother and more profitable sales operation. We touched on Shopify's AI first hiring earlier, but it looks like CEO Toby Littke is doubling down on integrating AI across the entire company. Yeah, Littke's latest directive really shows how serious Shopify is about becoming a truly AI powered organization.

He's basically saying, everyone use AI. It's not just an option anymore. It's expected and it's tied to performance evaluations and even hiring decisions. So it's not just about replacing jobs with AI. It's about every single person at Shopify using AI in their work. That's right. Shopify is providing a bunch of AI tools and setting up internal channels for employees to share tips and learn from each other. The message is loud and clear. AI skills are essential, not optional.

And as we talked about earlier, managers now have to explain why AI can't handle a task before they can even ask to hire someone new. Litke seems incredibly bullish on AI's potential to supercharge productivity. He's called AI a multiplier, something that lets top performers get way more done, even 100x the work in some cases.

This company-wide push shows just how much people believe AI can transform how businesses operate and innovate. And this isn't just Shopify. It feels like a broader trend, with more and more industries emphasizing AI skills. Absolutely. Shopify is taking a really comprehensive approach that could become a model for other companies looking to fully embrace AI to boost productivity and innovation. Okay, this next one is a bit of a curveball.

It's about AI and energy policy. I know, right? It's an unexpected connection. The White House is apparently using the growing energy demands of AI data centers as a reason to increase coal production.

Wow. So they're saying the electricity needed to run these massive AI systems is outstripping what our current renewable energy sources can handle. That's the argument. The energy appetite of these data centers is raising concerns about energy security and whether renewables can keep up with this rapidly grown demand. This has some pretty big implications for climate change efforts, doesn't it? Definitely. It's reigniting the debate about the environmental impact of large scale AI.

It forces us to ask, how can we balance the excitement of AI with the need to protect our planet? It's a really important issue to watch. Now, on a lighter note, let's talk about some exciting progress in AI voice technology. Amazon just announced something called NovaSonic. Ah, yes. NovaSonic is Amazon's new AI voice system designed to make AI conversations sound incredibly realistic.

The big breakthrough here is its ability to mimic human-like intonation and expression so interactions feel more natural and less robotic. So we're moving away from the monotone robotic voices we're used to with some AI assistance. Exactly. Novasonic is meant for conversational agents, virtual assistants, and even entertainment applications within AWS, Amazon's cloud platform. The goal is to make voice-based experiences richer and more engaging. Imagine being able to talk to technology.

And it feels like you're having a real conversation with another human. That's the potential. More natural, expressive voice interactions could revolutionize everything from customer service to education to how we create content.

And speaking of immersive experiences, Google's bringing AI magic to entertainment with a new Wizard of Oz show in Las Vegas. This is a super cool application of AI in the entertainment world. Google Cloud and Sphere Studios are teaming up to power the immersive experience using AI for all sorts of things.

Think stunning 3D visuals, voice processing, and even real-time scene generation. Hold on. Real-time scene generation? That sounds like the show can change on the fly. It can. This allows for unscripted interactions with characters and more spontaneous magical effects that can respond to the audience or the performers.

It creates a much more personalized, engaging experience for everyone. So AI is no longer just working behind the scenes. It's actively shaping the entertainment we consume and making it more interactive. Exactly. This collaboration represents a whole new frontier for AI and entertainment, blurring the lines between traditional storytelling and AI-driven experiences that can adapt and evolve based on the audience. All these advancements are amazing, but of course there are also some challenges emerging.

One that's particularly concerning is the rise of AI-driven fraud in hiring. Yeah, it's a worrying trend. Recruiters are seeing a huge spike in fake job applicants using AI to create flawless resumes, cover letters, and even deploy interview bots to impersonate real candidates.

So it's getting harder and harder for HR teams to tell the difference between real people and AI generated personas. Unfortunately, yes. These AI tools are getting really good at creating convincing application materials and even simulating impressive interview performances. It's a real challenge for companies trying to hire the right people.

It could be a huge problem for HR departments and online job platforms. It is. We need much better identity verification tools and more sophisticated fraud detection systems within hiring processes to combat this new breed of AI-powered deception. Okay, before we wrap up, let's do a quick rundown of some other notable AI happenings from April 8th, 2025. Sounds like it was a busy day in the world of AI. It was a whirlwind. As we mentioned earlier, Meta's GenAI lead strongly denied those benchmarking accusations about Lama 4.

Runway unveiled Gen 4 Turbo, a much faster version of their AI video generation model. Google expanded access to AI mode with its new multi-modal search to more users.

CREA, an AI-powered creative platform, scored a massive funding round to add audio and enterprise features. And there was the launch of a support-responsible AI campaign by a bunch of media outlets. Eleven labs rolled out a new MCP server integration for even better AI voice capabilities. Oh, and researchers at the University of Missouri showcased a cool new AI-powered wearable that monitors heart health.

Wow, that's a lot to pack into one day. It really shows how quickly this field is evolving. It's a testament to the incredible energy and the sheer scope of AI research and development happening right now. Now, if all this talk about AI has got you fired up, and you're not just interested in understanding but mastering the technologies behind it, then I have something you might like.

It's called the AI-powered Jamgak app, created by none other than Etienne Newman, the brains behind this podcast. This app can help you ace over 50 in-demand certifications in fields like cloud computing, finance, cybersecurity, healthcare, and business, just to name a few. It's like having a personal AI tutor right in your pocket. You can find the links to download the app in the show notes.

So let's recap our deep dive into AI on April 8th, 2025. We saw a ton of activity. There's the debate about transparency and AI benchmarks, big changes in the workplace driven by AI's growing capabilities, exciting advancements in multimodal AI applications, fierce competition for top AI talent, potential moves by big players like OpenAI into the hardware space, the expansion of real-time AI assistance, and the increasing automation of business processes. We also talked about how AI is intercepting with

unexpected areas like energy policy and entertainment, and the new challenges posed by AI-driven fraud and hiring. It makes you wonder, with all this rapid progress, what parts of our work in society will be most affected by AI in the coming years?

It's mind-boggling to think about the long-term impact. Absolutely. Staying informed is key in this fast-changing field. Hopefully, this deep dive gave you some valuable insights into the state of AI today. Don't forget to check out Etienne's AI-powered Jamcade app if you want to take your AI knowledge to the next level. All the links are in the show notes. Thanks for joining us for this fascinating deep dive into the world of AI.