Hardware’s Ongoing Affordability Crisis
It’s no secret that AI is driving up hardware costs. But what is less talked about is how it's impacting government buyers. Sure, they’re paying more for the same stuff they bought last year, but now, the slow pace of procurement is making it even more painful. Some estimates put cost increases at several hundred percent in very short timeframes.
An employee at one company in the hardware space recently called out the need to move faster on these critical buys or risk paying a premium for the basics. For government to be competitive and get the tools it needs at a price it can stomach, it needs to be nimbler and more adaptable to the markets, they argued.
The Next Public Records Crisis?
Government has another public records problem brewing with AI. Like the unsanctioned or unmanaged social media profiles that got T-boned by public records requests a decade or so ago, we’re hearing some skittishness around requests for AI prompts in state and local government.
Privately, some in the public sector have said that requests are ramping up dramatically and forcing policy changes in areas such as retention timelines. Imagine a monthly request for every prompt entered by every government employee within a large municipal government, as was the case with one source. That could be hundreds of thousands of records for large jurisdictions.
Last month, Industry Insider — California spoke with a First Amendment attorney who said there's a lot of grey area here, but that governments of all shapes and sizes should buckle up for the long haul. There are some questions about all of this, such as who actually owns the data queried through third-party models and whether these searches really classify as public records, etc.
For now, there will be a lot of catch-up to play, and these record requests could boil over into the courts as organizations push back.
Saltiness Over Price?
A sentiment echoed among some state officials lately is another cost concern — use pricing. Compared in one recent conversation to the surprise charges that accompanied early cellphone bills, the public sector is grappling with how to hit the moving budget target that is use-based.
Government, especially the state, is used to the capital outlay model of budgeting — five to 10 years out, predictable and safe. Industry, meanwhile, has moved on to technology pricing models that shake things up. That's prompting conversations that industry might not love in the long run; conversations such as “can we even afford to start down this road” or “maybe we should find another option.”
Public-sector folks are already making noise about reclaiming control for M&O projects, improving on the code they buy with AI’s help and unifying buying power.
On Thursday, Government Operations Agency Secretary Nick Maduros said during a Little Hoover Commission hearing that departments shouldn’t be paying different prices for the same tools and that the state should be leveraging its buying power more effectively.
Picking Poppy
The state has been diligently working on its internal AI assistant, Poppy, and it’s about to go mainstream. CDT’s Deputy State Chief Technology Officer Chaeny Emanavin said during a recent Code for America lightning talk that the tool was nearly ready for prime time.
The tool leverages in-house expertise and 10 large language models to assist state employees with various aspects of their daily work.
