The industry has a black box problem, says Big Blue
Today’s commercial AI models are opaque and unsuited for the enterprise – and adding your data doesn’t make them any safer.
In the latest episode of Computing’s Ctrl Alt Lead podcast, Leon Butler pulls no punches on the state – and future potential – of enterprise AI.
As businesses rush to deploy generative AI, IBM’s UK&I CEO warns that too many are pouring proprietary data into models they don’t fully understand. It is, he says, like dropping your own ingredients into an opaque test tube and hoping for a safe outcome.
“There’s a little bit of liquid in there that represents the data the model’s been trained on. You put your own liquid, your own data, in, you shake it up – but you’d never drink it because you don’t know what’s already in it.”
This episode tackles the real blockers for IT leaders: black box models, trust gaps, and the need for transparency before data fine-tuning.
Leon also describes how IBM is doubling down on automation through strategic acquisitions like Datastax and the WatsonX portfolio to offer AI that’s powerful, auditable and enterprise-ready.
And if you’re not thinking about quantum yet? You should be. Leon forecasts a 7% productivity boost for the UK economy this year, and urges organisations to start prepping for quantum-safe security now.
To access the full podcast watch the video, use the web player, or find Ctrl Alt Lead now on Spotify and Apple Podcasts.