You know that feeling when you're coding at work and reach for ChatGPT to debug something?

Well, Meta just said: "Why are we pretending that's not how real development works?"

Last week, they made a move that's got every tech recruiter scrambling.

Meta now lets candidates use AI during coding interviews.

The twist? They're not just handing over easy problems. They're giving candidates broken AI-generated code and asking them to fix it.

Some can be fixed. Some can't be fixed at all.

The real test? Can you tell the difference?

Why This Actually Makes Sense

Think about your last coding project.

Did you solve it by memorizing algorithms in a silent room with no internet?

Or did you use Stack Overflow, documentation, AI tools, and maybe ask a teammate for help?

Meta figured this out. Testing someone's ability to work with AI is probably more valuable than testing their ability to reverse a binary tree from memory.

It's like testing whether someone can cook a great meal with modern tools, not whether they can start a fire with sticks.

The Problem Nobody's Talking About

Here's the catch: How do you evaluate this fairly?

Traditional coding interviews were simple. Your code worked or it didn't.

Now you're judging:

• Can they spot bad AI suggestions?

• Do they know when to give up on an AI approach?

• Can they iterate effectively with AI tools?

Most interviewers aren't trained for this. It's like switching from grading a math test to judging an art contest.

The Security Challenge

If candidates can use AI, how do you make sure they're not getting help from their engineer friend from google?

The line between "approved AI assistance" and "unauthorized human help" just got very blurry.

Traditional security was easy: no internet, no phones, done.

Now? You need to verify the person is actually solving problems themselves, not just copying answers from somewhere else.

What Happens Next

Meta's move is already spreading. AI job postings jumped 20% this month.

The skills that matter are changing:

• Can you prompt AI well?

• Can you debug AI code?

• Can you explain why AI solutions won't work?

Smart companies will probably go hybrid:

• Some AI-assisted coding (like Meta)

• Some traditional problem solving

• Some collaborative discussion

The Bottom Line

Meta isn't just changing their hiring process.

They're acknowledging that great developers in 2025 need to be effective with AI as a tool.

The question isn't whether other companies will follow. They will.

The question is whether they can do it securely and fairly.

At Talview, we're helping companies navigate this shift with secure AI-powered proctoring that works for these new collaborative interview styles.

Because embracing AI in interviews is one thing.

Making sure the process stays fair and actually tests the right skills? That's the real challenge.

What do you think? Will this find better developers or just better AI prompt writers?