If you’ve been following the latest in AI, you’ve probably heard about OpenAI’s new o1 model. This model is designed to take its time when processing information, thinking more deeply before responding. It’s a concept that seems particularly well-suited to some of the problems we’re working on (i.e., the intricate world of public infrastructure projects). But let’s unpack how this might actually play out in the GovTech landscape, especially after what we heard at the recent State of GovTech conference.
At the conference, one thing was crystal clear: municipalities are increasingly looking to GenAI to help streamline processes. Things like enhancing community engagement with tools like Rep’d or fine-tuning models to help with administrative tasks were popular topics. The potential for these AI models to take over repetitive, time-consuming tasks is undeniably attractive no matter the sector. However, there was also a strong undercurrent of concern regarding cybersecurity and the responsible use of AI. These aren’t just buzzwords—they’re critical considerations for any organization looking to implement AI at scale, especially in the risk averse public sector.
This brings us to a crucial point: architecting software platforms that can fit the right model to the right task, while maintaining security protocols, will be a significant differentiator for companies like dependbuild (more to come on this later!). As we’ve seen with OpenAI’s o1, different models excel at different tasks. Some are great at reasoning through complex problems, while others might be more efficient or cost-effective for simpler, high-volume tasks. The ability to select the right model for the right task—balancing performance, cost, and the amount of context needed—could be a game-changer in the GovTech space.
For instance, OpenAI’s o1 model might be perfect for the complex, nuanced task of risk identification and management in infrastructure projects. These projects involve layers of regulations, safety standards, and financial implications that require more than just a quick AI response. However, it’s also important to consider the cost and computational resources involved. AI that takes more time to think also requires more processing power and, in turn, could become more expensive to deploy at scale. With ROI calculations being more abstract, this could easily price out the public market, especially at the local level. That’s where smart architecture comes into play—balancing these factors to deliver the best possible outcomes without breaking the bank.
Moreover, the integration of these AI models into platforms like dependbuild isn’t just about adding another tool to the toolkit. It’s about fundamentally enhancing how we operate. In our case, it could very well improve our risk assessment module. But, as with any new technology, there are risks and uncertainties.
For now, we’re doing our homework.