IT Essentials: Put up or shut up

Time for AI to prove its worth

After two years of experiments, AI needs to prove it can show measurable success.

We’re just days into 2025 and press releases about new AI-based products are pouring into my inbox – most from CES in Las Vegas. This is, after all, the year of the AI PC (so sayeth the marketers).

But AI in its current generative form has been around for more than two years, and it’s time – as Snowflake’s Jennifer Belissent told John Leonard this week – to show us the money.

At this point, AI trials are mature, and the technology’s abilities and limitations are understood. Now it’s time to scale those experiments, which is its own expensive challenge.

Return on investment has been a recurring theme at Computing events through 2023 and 2024, with a huge variety of metrics cited as measurement tools. That really demonstrates AI’s success in penetrating different areas of business, but also reflects some confusion about just how to measure success.

This is the year those metrics will need to firm up, with experiments that can’t prove their worth falling by the wayside.

The days of being able to win investment just because something has an AI label slapped on it are (slowly) fading. Not totally, though; some PR for an AI shopping assistant landed in my inbox as I was writing this. The hype isn’t over yet.