The FOMO-based frenzy over GenAI is continuing, with investment activity, marketing blitzes, conferences, product announcements and more all off the charts. We’ve seen this “leap before you look” excitement lead to bad outcomes so many times. Maybe this time around, we can get a bit of perspective, and have a plan, before the spending and development start.
To that end, I got together with my friend Shawn Rogers at BARC to do some real primary research on organizational readiness and the state of technology for the new AI stack. We surveyed 335 companies worldwide in February of this year – this is real data, and it’s new. We learned a lot. In this and a few future blog posts, I’ll highlight some of the results. You can get the whole 33 page study here.
One of the most important insights we gained came when we asked respondents about how ready they were. We asked about seven categories: Identifying Leadership; Security Standards; Data Access Policies; Program Policies; Legal Considerations; Project Governance; and Enterprise Architecture Requirements. We asked them to define what stage they were in for each: researching; consulting stakeholders; drafting policy; reviewing/revising policy; or formalized. (They could also answer Not under discussion – roughly 5% did.)
The results were fascinating – the details are worth digging into when you get your hands on the research, but I want to highlight a key finding – some 20.5% of the participants can be characterized as “High Readiness.” That means they were in the top two stages across all seven categories. And having seen this, we went back through all the data again, using High Readiness as a filter to see how they differed from the general population in the survey. What that analysis taught is this: there are clear lessons to be learned from their experience. This may seem intuitively obvious, but don’t assume. One way to identify pioneers is often to look at the arrows in their carcasses.
I’ll just highlight one finding here:
High Readiness companies are all in on security and compliance. And you should be too.
The big “AI failure” stories that are inevitably coming will no doubt be centered around these issues – opening up your precious data to make your new systems smarter and minimize hallucinations is imperative – but you have to do it right. Get those standards and policies ready first, not after the breaches become visible.
I’ll highlight a few other findings in future posts. But you don’t have to wait. Get your copy of the study now. The data is fresh, and much of the messaging you are seeing now, when it is data-based at all, is old. Things are indeed moving fast. Arm yourself with fresh data.