🔥 Can AI Power Up Without Burning the Planet?

In today’s email:

🌍 AI's carbon footprint is growing right alongside its power demands—and the tech giants are scrambling for a solution.

🤖 AI might be able to empathize, but it's still working through some serious bias issues, especially when it comes to race.

🚚 Logistics are getting a major upgrade, with AI leading the way in optimizing everything from routes to real-time tracking.

Intrigued? Keep on scrolling!

🌎 AI Models Are Growing—And So Is Their Carbon Footprint

AI’s energy needs are spiraling out of control, and it’s getting awkward for tech companies. You know those boring, windowless buildings—data centers? Turns out they’re like chain-smoking chimneys, quietly puffing out tons of greenhouse gases while fueling our AI obsession.

Here’s the situation in bullet points:

• Carbon emissions from data centers have tripled since 2018—they’re now right behind airplanes in pollution rankings.

• AI models keep getting bigger and hungrier (thanks, video tools like OpenAI’s Sora).

• Most data centers are in coal-heavy regions, making their energy 48% dirtier than the national average.

So what’s the plan? Nuclear power is back on the table. Meta, Microsoft, and Amazon are all chasing nuclear partnerships, but nuclear plants are slow to build, and public opinion is lukewarm at best. Meanwhile, AI companies are eyeing Southeast Asia as their next energy playground.

The bottom line? AI companies are stuck between their sustainability promises and the race for smarter, faster models. Until they figure it out, data centers will keep slurping up power—and spitting out carbon. So, next time you see one of those plain buildings, give it a little side-eye. It’s working way harder than it looks.

🐞 AI’s Empathy Has a Bias Bug

So, you know those mental health chatbots popping up everywhere? Turns out, researchers from MIT, NYU, and UCLA decided to put them under the microscope to see if they could actually dish out helpful, empathetic advice. Using a boatload of Reddit posts (12,513 to be exact), they compared GPT-4 responses to human ones. Here’s what they found:

• GPT-4 is surprisingly good at this whole empathy thing!

  • AI responses were 48% better at encouraging positive changes than humans.

• BUT—and it’s a big but—AI’s empathy drops when race is involved:

  • Black posters: 2-15% less empathetic responses.

  • Asian posters: 5-17% less empathetic responses.

• Humans also showed bias, but more subtly—acting nicer when hints about race were implied rather than explicitly stated.

The good news? Researchers discovered that telling GPT-4 to account for demographics upfront actually reduces bias. Who knew instructions mattered so much?

While AI therapists are promising, they’re not perfect yet—especially for marginalized groups. AI may have a big heart, but it still needs a little tuning to play fair.

🚚 AI: The New Brain of Logistics

AI is shaking up the logistics world in some big ways, helping companies stay afloat during tough times. Between the Suez and Panama Canals taking a nosedive in trade volume, delivery times are dragging, and inventories are struggling. AI is swooping in to save the day.

Here’s a quick rundown of the AI innovations that are shaking things up in logistics:

• AI-Enabled Vision Systems: Think 3D cameras and smart software helping trucks and warehouses track everything—road conditions, product damage, stock levels—basically, it’s like giving planners an extra pair of supercharged eyes.

• Real-Time KPI Tracking: AI tools analyze truck data, weather, and fuel to optimize routes and boost efficiency. Plus, these systems help with training drivers and offering maintenance tips, all in real-time.

• Optimized Planning With Composite AI: This AI does a bit of everything—network design, procurement, route planning, you name it. It mixes techniques to make logistics operations as smooth as possible, keeping costs low and meeting demand efficiently.

Other cool AI stuff that is trending right now 🔥🔥

🌍 AI might be the future, but if we keep ignoring Africa’s languages, talent, and tech gaps, we’re basically asking a supercomputer to solve global problems with half the world’s data missing. - Read more

🏠 Rejected by an AI landlord, Mary Louis fought back, sued, and scored a win that might just save renters like you from getting ghosted by a glitchy algorithm. - Read more

📞 AI voice scams are getting slicker, but don’t panic—creating a family "safe word" could be your best defense against these high-tech con artists. - Read more

👓 Meta’s smart glasses just got a major upgrade with live AI, real-time translations, and Shazam—because why not have your glasses help you chat, translate, and ID songs all at once? - Read more

▶️ YouTube is letting creators finally call the shots on whether AI companies can use their videos to train models. - Read more