AI is Confidently Wrong

Hello,

This is Simon with the latest edition of The Weekly. In these updates, I share key AI related stories from this week's news, list upcoming events, and share any longer form articles posted on the website.

My CEO has been in our London office this week, and during a presentation, said something that stuck with me.

"AI can be confidently wrong."

We’ve spent the best part of our careers with tools that have very clearly shown errors. If a formula in Excel breaks, you see an error message. If a search comes up empty, it tells you. AI has been immediately different. It gives you smooth, well-written answers, even when they’re totally wrong. And this is a real issue.

The mistakes nobody talks about

When people talk about AI mistakes, they usually talk about news worthy stories like deepfakes, job losses, and poor policing decisions. But today, I want to talk about the quiet mistakes, the ones that slip through during a normal working week.

Maybe it’s a statistic in your slides that’s almost right but not exact. Or a meeting summary that covers the discussion but skips the real decision. It could be an AI-drafted client email that sounds okay but feels off, or a contract summary that quietly misstates a clause because no one double-checked.

These mistakes don’t make headlines, but they happen every day in fast-moving businesses that rely on their tools.

Why they get through

The reason these errors are so hard to catch is the same reason AI is so useful: it sounds authoritative. The output is coherent, grammatically correct, and presented without hesitation.

When a coworker isn’t sure, they’ll say things like "I think" or "maybe double-check this." AI doesn’t do that. We read what it gives us, it looks fine, and we move on.

The mistake usually shows up later.

This isn't an argument against using AI

Plain AI isn’t about avoiding these tools. It’s about using them with your eyes open.

There’s a simple but important shift: instead of asking "is this good enough?" try asking "where could this go wrong?"

For facts, double-check numbers, dates, and sources, especially if it’s going outside your team. For summaries, ask if what’s missing is actually the most important part. For messages, think about how the tone will feel to the reader, not just what the words say.

AI is great for a first draft, but the final judgement is still up to you.

One thing to try this week

Before you send, submit, or publish anything made with AI, ask yourself: what’s the one thing here that would be most embarrassing to get wrong? Check that one thing. It only takes a minute, and it’s the difference between AI being a risk and AI making you better.

The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.

Learn more

This week that Microsoft, Chevron and Engine No. 1 signed an exclusivity agreement tied to power generation for data centres, as tech companies race to secure electricity for AI workloads. Microsoft President Brad Smith said local community support is now critical for building data centres, because towns are pushing back over rising power prices, water use, and pollution concerns.

Why it matters: AI adoption is no longer just a software story. Energy supply, local planning, and infrastructure bottlenecks are becoming strategic business issues.

Oracle said it is reworking its cloud software suite into what it calls “agentic apps”, designed so companies can ask for business outcomes rather than manually handle every workflow step. Reuters said the changes affect software used for major functions such as production planning, finance, customer collections, and HR, with AI taking on more of the execution while humans focus on judgement and trade-offs.

Why it matters: This is the clearest sign that AI is moving from standalone assistants into the systems businesses already run on.

Legal AI company Harvey raised $200 million this week at an $11 billion valuation, with Reuters reporting that the money will help expand its AI agent and legal engineering teams. The company’s tools are used for work such as contract analysis, due diligence, compliance, and litigation.

Why it matters: One of the most commercially meaningful AI trends is the move into specialist, high-value workflows rather than generic chat tools.

Thanks for reading, and see you next Friday.

Simon,

Was this email shared with you? If so subscribe here to get your own edition every Friday.

Enjoying Plain AI? Share it and get a free gift!

If you find this newsletter useful, why not share it with a couple of friends or colleagues who would also benefit? As a special thank you, when two people subscribe using your unique referral link, I’ll send you my "Exclusive Guide: Supercharge Your Work with NotebookLM." It’s a practical, no-nonsense guide to help you turn information overload into your secret weapon.

You can also follow me on social media: