AI and bureaucratic productivity
« previous post | next post »
The August 6 Economist discussed AI threats to government bureaucracy:
- "How AI is breaking the British state: Equipped with models and agents, citizens could bring government to a halt"
- "From parking tickets to planning, the state is under siege by AI: The people-versus-state arms race"
- "The tragedy of the commons, AI edition: Britain’s employment courts are clogged with AI cases"
On August 7, Henry Farrell weighed in, with "The downside of robot solutionism: Mapping the Abelard Snazz feedback loop" (see here for an Abelard Snazz summary). And on August 10, Cory Doctorow added "The bureaucratic arms-race is mutually assured destruction".
These articles are about how new kinds of automation can paradoxically decrease the productivity of bureaucracy by increasing the productivity of submissions, denials, and appeals.
I discussed a different dimension of the problem in "What AI is (not) good for?" (8/4/2024):
[T]he main effect of several such out-sourced innovations in my own workplace has been to force many departments and centers to hire new human employees to deal with the bureaucratic "help" thus provided, which semi-automatically monitors lots of newly-tracked details through badly-designed interfaces to ill-fitting data models. Which is not a very good imitation of the 19th-century successes of industrial automation…
This seems to be a problem with the people designing, selling, and buying such systems, rather than with the underlying technologies. But as long as administrative productivity is defined by bureaucrats, I don't think it will be easy to fix the problems.
From David Segal, "So, Human Resources Is Making You Miserable?", NYT 8/3/2024:
A recent Business Insider story, titled “Everyone Hates Workday,” asked why half of the Fortune 500 uses this particular piece of software — which handles benefits and recruiting and facilitates pay equity analysis — despite creating “mountains of busywork for everyone.” […]
WorkDay's web site boasts
At Workday, AI is at the core of our platform. And as part of your workflow, it powers intelligent predictions and automation like no one else can.
As the Business Insider story explains — and as everyone I know with WorkDay Experience can testify — it also creates layers of counter-productive frustration.
See also:
"A dangerous degree of accidental intelligence", 10/1/2023
"Macroeconomics of AI", 4/23/2024
"AI: Not taking jobs yet?", 10/2/2025
Update — Callie Holderman, "Chatbots Are Pushing Us Toward a Post-Human Internet", NYT Magazine 8/14/2026.
AntC said,
August 16, 2026 @ 5:12 pm
There's a longer-standing and (arguably) nobler Information Technology term 'Business Intelligence' — which is 'intelligence' more in the sense of actionable information (compare 'military intel') than 'capacity of mind' [as wiktionary puts it].
It seems a lot of BI vendors have rebranded themselves as AI, to jump on the bandwagon. 'Workday' is a Peoplesoft-alike BI offering, rebranded. (Compare how 'e-business' rapidly became a rebranding of 'we have a website'.)
Mark Liberman said,
August 17, 2026 @ 5:18 am
@AntC:
The near-term effect of "AI" on jobs is mainly through the (further) automation of things like "Business Intelligence" — human resources, financial management, marketing and sales, customer relations, licensing and permits, etc. This automation process has been going on for decades, but the rate and extent have sharply increased, under the flag of "AI", facilitated by massive growth in cloud computing.
And so far, the main results of this process seem to be (1) increasing hatred of its authors and actions, and (2) major growth in bureaucratic slopwork, much of which may be due to the Abelard Snazz feedback loop, but meanwhile also requires more human labor rather than less, in order to maintain actual function in actual organizations.
Andrew Usher said,
August 17, 2026 @ 9:47 pm
Funny that the Economist articles that started this post – all three – were about the common people trying to assert some rights against a system they're relatively powerless against. I suppose it makes sense that that's the only problem with AI to the establishment mind. Use by big business (and big government)? – Naturally OK (we don't have to think of what it means for people, it's just the inherent 'need' for bureaucracy to keep up with the times and endlessly change while staying the same, and its unacknowledged 'need' to dehumanise people while allowing those in charge to think they're not.
These kind of people would find the idea of an AI arms race – far from hypothetical at this point and the only really new thing related to AI – incomprehensible. Until it happens, of course, then it's 'not our fault'.