New Graduates Are Getting Hired Despite AI Fears — And the Data Is Starting to Show Why
Something interesting is happening in the world of AI right now: the predictions aren’t matching reality. Entry-level jobs were supposed to be disappearing. Big energy bets are falling apart. And one of Meta’s quieter tools is suddenly getting a spotlight. Today’s stories reveal how messy and unpredictable AI’s real-world rollout actually is.
The AI Jobs Apocalypse for New Grads Hasn’t Arrived Yet
For the past two years, economists, headlines, and nervous parents have been sounding the alarm: AI would gut entry-level work. Interns, junior analysts, and recent graduates would be the first casualties. So far, according to Ars Technica, the unemployment numbers aren’t cooperating with that story.
Recent graduates are finding jobs at rates comparable to their predecessors before AI tools became widespread. That doesn’t mean AI has had zero effect on hiring — companies may be doing more with fewer people overall — but the specific collapse in entry-level roles that many predicted hasn’t shown up in the data yet.
Why might that be? One theory is that AI tools have actually made junior employees more productive, which makes them more valuable to hire, not less. A new grad who can use AI to do the work of someone two years more experienced is still worth bringing on. Replacing humans entirely turns out to be harder than automating pieces of their jobs.
For anyone anxious about entering the workforce, this is genuinely reassuring — with a caveat. “So far” is doing a lot of work in that sentence. The data reflects a moment in time, and AI capabilities are still advancing quickly.
Why this matters: The fear of AI eliminating early careers may be causing unnecessary panic. But watching these numbers over the next few hiring cycles will tell us far more than any single snapshot.
“New graduates still finding work at comparable rates to previous years.”
A $1.25 Billion Energy Bet Just Collapsed
AI data centers — the large facilities packed with specialized computers that power tools like ChatGPT — consume enormous amounts of electricity. Finding creative ways to power them has become one of the industry’s biggest challenges. Crusoe Energy thought it had a bold answer: partner with Boom Supersonic to use experimental stationary power plants, essentially jet-engine-derived turbines repurposed to generate electricity on-site.
That plan is now reportedly dead. According to TechCrunch, Crusoe has walked away from the $1.25 billion arrangement. The most likely explanation is that Boom’s technology simply isn’t ready to deploy at commercial scale on the timeline that Crusoe needed.
Think of it this way: you’re building a restaurant and you’ve promised your investors the kitchen will use a custom oven that doesn’t quite exist yet. At some point, you have to order a regular oven or you’ll never open.
For people who use AI tools daily, this is a reminder that the infrastructure behind those tools is genuinely fragile and expensive. Every time you run a query, something is burning fuel somewhere. Finding stable, affordable power is a real constraint on how fast AI can grow — and ambitious experimental solutions don’t always pan out.
Why this matters: The AI industry’s hunger for power is running into the limits of what today’s energy technology can reliably deliver. Expect more partnerships to be renegotiated or scrapped as that tension continues.
“Crusoe abandons $1.25B plan to use Boom turbines at AI data centers.”
Meta Opens a Waiting List for New Muse Features
Meta’s Muse — a creative AI tool the company has been developing — is reportedly expanding, according to TechCrunch. Meta is now accepting signups for an early access program, letting interested users join a waiting list to try new capabilities before they roll out more broadly.
Details are thin. Meta hasn’t publicly described what the new features do or when most users will see them. Early access programs like this are typically how companies test new functionality with a smaller group before a wider release — gathering feedback and catching problems without risking a chaotic launch.
For people who use Muse creatively, or who are curious about Meta’s AI direction, signing up costs nothing but time. Whether the new features represent a meaningful upgrade or incremental polish isn’t clear yet.
Why this matters: Meta is making a quiet push into creative AI tools at the same time it’s loudly promoting smart glasses. Muse could become more significant than it currently appears.
“Meta launches early access program for Muse features via waiting list.”
Also Happening in AI
On the developer side, OpenAI quietly released version 3.19.1 of its Python library — Python being the programming language most commonly used to build AI-powered applications — fixing a bug in how its chat feature handled certain inputs. LangChain, a toolkit that helps developers connect AI models to outside data and tools, also pushed a minor update to its Fireworks integration. Over at The Verge, a hands-on comparison found Apple’s new AI-powered home camera features holding their own against similar offerings from Amazon and Google. Meta’s Connect conference is reinforcing a clear bet on smart glasses as an everyday AI device. And TechCrunch flagged a concerning security story: some users of Supabase, a database service popular with developers, have accidentally left their databases wide open to the internet, exposing private user data.
What to Watch
The graduate employment numbers deserve a longer look — the real test comes when this year’s hiring cycle closes in late fall, and whether AI-related cost-cutting finally shows up in entry-level headcounts. On the energy side, watch which power sources AI companies pivot toward now that experimental deals like Crusoe’s are collapsing; nuclear and grid-scale batteries are the leading alternatives, and announcements in that space could come fast. Meta Connect’s emphasis on smart glasses also signals something worth tracking: the next wave of AI may not live on your phone screen at all.