Over the last few years, the use of generative AI has dramatically accelerated as cities, counties and state governments use large language model tools for document summary and translation, data analytics and software code generation.

But according to a March report, “The Rise of Agentic AI in State Government” from the National Association of State CIOs (NASCIO), there’s another giant step on the horizon: Agentic AI. 

Among other things, agentic AI can move toward multistep execution without human integration. And that’s just the beginning of the list.

The following points, drawn from the NASCIO report, helps to explain the potential advantages of agentic AI. It: 

• Has persistent memory, which remembers past interactions and adjusts plans accordingly while GenAi has only short-term context recollection.

• Is able to move forward with minimal human input while Gen AI uses human prompts to generate responses.

• Evolves learning from interactions which allow refinements to behaviors, while Gen AI is static and learns only by retraining from developers. 

• Drives automation, reduces human workload and enhances business capability with a time savings of between 30 and 60 percent, which gen AI, on average increases speed up of activities by up to 25 percent. 

This all sounds like a wonderland of advanced technology, but the NASCIO report counsels that this evolution will be gradual. As the report explains, “State technology leaders are unlikely to walk into the office one day and say ‘This is the day we start using agentic AI.’ It’s more likely that GenAI tools are starting to add capabilities, extending existing GenAI tools incrementally while keeping humans in the loop. Eventually, states will have crossed in to ‘agentic’ territory. In some cases, perhaps the tool itself hasn’t changed but the responsibility the tool has been allowed to take on has.”

A few of the states that are in leadership positions in making this transition include Alaska, Tennessee, Delaware and Texas.

Of course, while many states have moved quickly to try out GenAI, “giving AI agents autonomy is something that may need to develop more slowly. States are faced with realities of governance, risk, legacy systems and workforce resistance to change,” NASCIO reports.

#StateandLocalManagementAgenticAI #StateTechnologyManagement #StateEarlyEffortsInAgenticAI #StateGovernmentGenerativeAI #StateArtificialIntelligenceManagement #StateArtificialIntelligenceDecisionMaking #StateandLocalAIGovernance #StateManagementAgenticAI #StateMovementToAgenticAI #AdvantagesOfAgenticAIforStates #AIandHumanResponsibility #AgenticAIandPersistentMemory #AgenticAIandEvolvedLearning #AgenticAIandAutomation #AIandHumansInTheLoop #BarrettandGreeneInc