A cautionary tale for AI usage: 100% turnover two years in a row
I wanted to post this here as I thought it would be relevant.
A firm I previously worked with, had the CEO go 100% on AI since aroundredacted% of staff quit in 2025 and again 100% of staff have quit inredactedIn 2024, the CEO basically just started inputting all emails and work product into GPT and all my feedback was from GPT vs CEO.
The business has two parts to it: leasing, and professional services. Org was doing about 2.2M in revenue with lease rates at about 90% inredactedLeasing has dropped to about 50% with most previous lessees having left due to quality issues, which means professional services revenue have dropped as well. The firm has lost enough cash flows that they are actively cutting hours of part-time workers.
Here are some highlights of where AI screwed up the most:
1) pricing for repairs/upgrades: AI had given the CEO fictional numbers, way below market norms, to use in budgeting repairs/upgrades.
2) creating work expectations not grounded in any reality. CEO asked AI to come up with work for various departments. AI gave them deliverables for employees to do. Employees realized the CEO was asking for the impossible, so they all used AI to give CEO the work AI had prescribed, only for the CEO to get pissed that it was AI slop. The majority of the work would require weeks/months of offline work, which AI assumed could be done in a week.
3) creative accounting. Experienced accountants do one thing, AI does another thing, leadership believes AI. Think changing depreciation and inventory calculations ad-hoc so there is no consistency between months/years, but the overall numbers look good.
I thought this would be useful to the community as a reminder to scrutinize AI output. I do use AI on a near daily basis, but I know the limitations.