AI is advancing fast—but your people and oversight still matter most

Big moves bring both opportunity and questions

This week, M&S committed £340 million to an automated warehouse in Daventry. At the same time, TikTok turned to AI for moderation and the UK government considered rolling out ChatGPT Plus—though the cost proved too high. These stories show AI’s power—and that its value depends on how we use it. 

Efficiency is tempting—but reliability counts

Automated systems, like at M&S, offer speed and scale. They promise smarter, faster fulfilment. But they cost time to implement and oversee. You still need people who know the process—and who can pause or adjust when something goes off chart.

AI needs governance—not just deployment

TikTok’s shift to AI moderation raises a warning: AI can’t judge nuance or user sentiment on its own. For your business, whether you've got a chatbot or auto‑sorting system, make sure humans review outputs, especially early on. Log what the AI does and why. That helps you improve—or stop something before it’s a real problem.

Value needs real human insight, not hype

Discussions around national ChatGPT access show the appeal of ubiquitous AI—yet without clear benefit or budget, it falters. For small businesses, that means ask “What will this cost—and what will it actually do for my team or customers?” Build in measurement from day one.

Three steps to use AI that works—and lasts

1. Choose a single task to pilot — Let AI assist, not replace. Maybe sorting emails, drafting content or flagging customer messages. Review everything.

2. Track outcomes clearly — Time saved, errors avoided, or happier customer replies. If it’s improving things, expand gradually.

3. Keep human control at the centre — AI outputs should always be checked. Log decisions, allow feedback, update how prompts or rules work.

   
   

Three practical takeaways

• Test one thing carefully—pilot, measure, then expand if it adds value.

• Build oversight into the process—AI needs rules and review to work well.

• Think cost and benefit clearly—AI that doesn’t earn its keep isn’t worth it.