Why FP&A Needs Human Intelligence in AI

‍ChatGPT passed the Turing Test a month ago.

So are you ready to ask it what your budget should be?

The Turing Test evaluates an AI model’s ability to exhibit human behavior. In April, ChatGPT 4.5 convinced testers it was human 73% of the time - more than actual humans who only did it half the time!

But mimicking human behavior - picking up on nuances, understanding humor, etc. - is one thing. Prescribing how businesses should budget spending, make capital investments, or pricing offerings is another thing.

That’s NOT to say you shouldn’t trust AI or FP&A software with AI features. Nearly all finance software comes with AI features bolted on, and some even have AI built into their cores.

AI definitely adds value and efficiency to the planning process, particularly when it comes to things like:

• Using historical patterns to perform predictive forecasting
• Providing non-finance users (operational teams) with natural language interactions
• Include operational data to better align fragmented systems for better business visibility

But even with AI some direct human involvement is needed, your FP&A process should use “human in the loop” AI-driven software.

Humans - you and your team - are the only ones who can verify that:

• A given correlation or pattern is valid, relevant and that you can act on it
• The right data is always used for plans and forecasts.
• You’re automating the right tasks, and a human is the one doing the approving.

So do trust AI based FP&A software. Do trust it for all the routine, repetitive, time sucking work you and your team don’t have time for - and don’t do well.

Back when I was working with MindBridge we always said “AI won’t replace Auditors, but Auditors using AI will replace those that don’t”.

For FP&A teams today, I would say the very same thing.

PS…for more on GPT passing the Turing Test, check out: https://lnkd.in/eDXZ6Q7w