How does AI contribute to your team during an incident?
In context to people risk technology (as with much of enterprise tech) as we introduce more AI capabilities, such as chat ... does the customer become the product?
The move to building an organisation's people risk data into an architecture where we can use AI to analyse, create, enact and scrutinise (whether agentic or generative) ... has this shifted the capability of 'what the user needs to know or do' from the products/tools we have today, onto the competence of the user?
So, let's assume you now have much or all of the data you need at your finger tips to understand the depth and impact of an incident, when planning policy or to analytically critique where exposure might reside. With chat interfaces that we are familiar with today ... that moves the ability to articulate needs onto the user. Are they good at that? Are we turning all our users into prompt engineers? Does this help or hinder them in a high pressure incident?
An interesting challenge to all of the technology providers is ... longer term, beyond building that data architecture ... if the users are competent in defining/building the tools for what they need from the AI toolset ... what are you bringing to the table?
In the people risk technologies we use today, the vendor has a deep understanding to why features exist, how they work and what those outcomes need to be. As a user, do I really want to define everything I want/need in chat prompts ... or do I need somebody who really understands all of that to build the tools for me? Its almost as if you could convert a highly detailed RFP automatically into a platform. I don't think we have removed the need for our risk vendors, their products, their features and strengths ... but we seem to be in this middle ground where we let the users define some aspects of what they want for themselves (reports, situation summaries, assessments, compliancy gaps etc.). It does put a responsibility on those users to do their homework, be good prompt engineers, check the reasoning, citations and outcomes.
At some point, if you look at most questions/asks users make of the AI platform to collate, enact, summarise etc. ... do you reverse engineer those needs back into your product and ensure those are tools or features you provide out of the box, perhaps proactively acted upon before somebody even asks their questions? Something to chew on for the product managers out there.
Comments