Generative AI’s Slowdown is an Opportunity, Not an Alarm
Generative AI’s rapid rise over the past few years has sparked both excitement and skepticism. The recent slowdown in the advancement of large language models (LLMs) might be seen by some as proof that the technology was overhyped. Make no mistake, there has been plenty of hype, but there’s also real promise. In my last post, I argued that cities could afford to “slow their roll” on generative AI fast adoption. The technology is still in its early days, and there is ample time to get it right. As it turns out, cities didn’t need to slow down — the AI industry itself is now hitting technological limits.
In a recent Bloomberg News article, Parmy Olson aptly describes this slowdown as a chance to catch our breath. Amid the rapid pace of model releases (and relentless hype), it often feels like a race just to keep up with generative AI, with no clear finish line. It’s been a particularly challenging time to demonstrate ROI when models are constantly being updated, and operational costs remain opaque. For local governments, this pause couldn’t have come at a better time. Organizations are learning how to address data privacy concerns, a lack of enterprise-level access controls, and the immaturity of many tools — but this knowledge is not widespread — it’s nascent and evolving. A temporary plateau in the state of generative AI gives more local governments the opportunity to plan and test solutions, and not rush into the unknown, unprepared.
The slowdown also provides policymakers with a much-needed buffer to establish thoughtful safeguards for secure and responsible use of generative AI. Where governments have been tasked with adopting generative AI, they’ve had to wrestle with a whole new, and ever-changing technology and to adapt current risk frameworks to encompass it. Using this time to ensure quality-controlled data and clear-eyed risk assessments will help avoid high-profile failures that could erode public confidence in all AI initiatives.
This is a positive development in my view. It’s not a warning sign that generative AI is going to be like the metaverse. We’re on the S-curve of technological adoption, where early investments take time to translate into clear value. Local governments should embrace this period to be deliberate and strategic. It gives local governments the time to shore up and expand their data governance, cyber, compliance, and risk assessments to accommodate generative AI. This is especially needed for the AI capabilities that are being dropped into existing govtech platforms.
