I've been building software since before "AI" meant anything more than a chess program cheating at knight forks, and I've learned to be suspicious of any headline that says a technology is about to hollow out the workforce. So I was pleased to see Ars Technica's write-up of Google's own analysis of 15 million real-world AI interactions. The conclusion, in plain terms: most jobs remain largely untouched. People are using AI to help with tasks, not to replace themselves wholesale.
This matters enormously for anyone in the automation and AI-marketing space, myself included. It's a useful corrective in both directions.
The hype has outrun the reality
If you've spent any time on LinkedIn in the last two years, you'll have seen the confident predictions: entire departments gone, copywriters redundant, developers obsolete by Christmas. Google's data — drawn from actual usage, not survey guesses or vendor marketing — tells a much duller story. People ask AI to draft something, summarise something, or unstick a piece of thinking. Then they get on with their job.
That's not a failure of the technology. It's just what adoption curves actually look like when you strip away the noise. I've watched three or four of these cycles now — desktop publishing, the early internet, mobile, cloud — and the pattern is always the same. The tech is real, the timeline is not what the hype merchants promised, and the winners are the people who build boring, reliable tools around the genuine use case rather than chasing the imaginary one.
Where the value actually sits
This is exactly the terrain I've built the Masher suite for. RSSMasher, MarketMasher, Article2Video — none of them promise to replace a marketing team. What they do is take the repetitive, high-volume, low-judgment work — turning a feed into a draft, a book into marketing copy, an article into a video script — and hand it to a human who then makes the actual decisions. That's augmentation, not replacement, and it turns out that's precisely where Google's data says people are using AI too.
If you're building or selling automation tools, this is worth sitting with. The pitch that resonates isn't "AI will do your job for you." It's "AI will do the 40% of your job that's mechanical, so you can spend your time on the 60% that actually needs a human brain." That's a much smaller promise than the hype cycle wants, but it's the one people actually act on, and it's the one that survives contact with a real budget conversation.
Don't underestimate it either
The flip side of this reality check is just as important. Some people, having watched the hype deflate a little, will now swing the other way and decide AI is overrated, a novelty, nothing to build a business around. That's just as wrong as the "everyone's job is doomed" take. The data doesn't say AI isn't useful — it says AI isn't yet replacing entire roles. Those are very different claims.
Fifteen million interactions producing real, measurable time savings on drafting, research, and content production is not nothing. It's the same shape of change that spreadsheets brought to accounting departments in the eighties — nobody got fired en masse, but the job of "accountant" changed permanently, and the firms that adopted early had a durable edge for a decade.
The practical takeaway
If you're marketing an AI product right now, calibrate your promises to what Google's own numbers show: augmentation, task-level efficiency, meaningful but bounded. If you're a business owner deciding whether to invest in automation, the sensible move isn't "wait and see" or "go all in" — it's to find the specific, repetitive tasks in your operation that are genuinely mechanical, and automate exactly those, while leaving judgment-heavy work to people.
That's not a thrilling story to tell on stage at a conference. But it's the one backed by 15 million data points rather than a keynote slide, and after forty years of watching technology promises come and go, I know which one I'd rather build a business on.
— Wayne