[JohnMakin]

> For ongoing projects, an effective trick that I believe I picked up from Secrets of Consulting is the anonymous poll, where you can ask individuals to rate their opinion of an AI project’s success chances on a scale of 1 to 10. The typical split I have observed is half of those involved rating the project at a 3/10 and others at around an 8/10 – a clear bimodal split on a project that was already three years late. Bringing this data to a CEO can be an effective method of pointing out that some information is clearly being hidden from them on the state of the project.

I don't think this is a very effective method of getting any kind of truth. Just as the people pushing the initiatives up top have incentive to lie about its success, people on the ground have incentive to lie about the opposite - for instance, in my org, we've pretty much just used it to automate tedium, and accelerate processes, and realized pretty quickly there were roles that were almost entirely automatable tedium and managing process - surprise surprise, those people are very negative about how useful AI is.

The tools are useful, there is no doubt, but the excitement and investment is simply way, way too high. I'm lucky at my org they have taken a more cautious "let's see what happens" approach, and want proof of any claimed successes.

[trescenzi]

Yea we saw an anonymous question “Are you using AI to be more efficient” start at 30/70 yes/no and flip to 90/10 after a mandate. I know that’s a very different question but it shows that anon questions are still directly influenced by everything else going on. Hopefully a question like the author recommends would be more accurate but I’d still expect a heavy influence based upon management’s stance.

[simoncion]

Savvy folks know that anonymous surveys that require you to sign in with any account are not anonymous. If a director -say- knows how many people in which groups have and have not filled out the survey, it's very much not anonymous.

[inigyou]

So is any anonymous survey that you got emailed a link to. That link is certainly different from your coworker. There are far too many stories of "anonymous" survey answers being used to discipline employees for anyone to trust the any more.

[simoncion]

Yep.

The only surveys you can trust even a little are the ones whose link gets sent to a wide-spread mailing list and work just fine without presenting anything tying you to the company.