Is Your Firm Wondering How It Should Move Forward With AI?

1. Task IT with creating a handful of fully functional, lighthouse implementations

The cost of these is dropping dramatically, as is the timeframe for execution. Frequently, we see them marrying more traditional algorithmic AI with generative AI. That aids with controlling the outputs.

IT may be the best department to work with private data clouds and other tools that will become essential as the tech spreads. But don’t assume they’ll be terribly user-centric in implementation. Limiting the number of IT-driven forays can ensure that each benefit from meaningful user input and isn’t simply tech-push. It also ensures that you’ll do these well.

2. Empower departments to undertake their own experiments, with very clear guardrails

(e.g. AI can summarize customers’ unstructured data or inputs, but not provide them with unedited outputs).

Be careful, though – just as IT-led ventures can sometimes become a low utility Science Fair, unsupervised experiments can become Pilot Hell. Insist that these experiments have clear hypotheses, measurable outputs, and specific mechanisms to capture learnings and iterate approaches. Experimentation isn’t a core competency of many companies outside their R&D departments, and the technological revolution that’s upon us is going to increase the need for experimentation quite rapidly.