Building on New Jersey's longstanding experience using new technology to engage with the public in how we make policy, the State AI Task Force pioneered a novel approach using AI to help us develop more robust recommendations faster and with the benefit of large-scale community engagement.
Rather than relying solely on traditional bench research or expert consultations, the Task Force's Workforce Training and Jobs of the Future Working Group developed a process that paired AI-powered research with direct input from thousands of New Jersey workers to address the pressing challenge of the impact of AI on work.
As a result of this process, which the Working Group undertook over eight weeks, New Jersey is implementing free AI skills training for all public servants and developing an AI-powered labor market monitoring system to help workers navigate career trends.
The Working Group was tasked with understanding how artificial intelligence will shape the future of work and how the State government, education institutions, unions, companies and other organizations in New Jersey can develop pathways to careers in AI – a large and complex topic with many unknowns. The challenge was significant: how to rapidly analyze vast amounts of evidence about AI's impact on work while ensuring recommendations reflected the real needs and concerns of New Jersey's workers.
Traditional policy research approaches often face trade-offs between the breadth of evidence that researchers can review and depth of community input that policymakers can gather. Too often, policies are developed without meaningful input from affected communities. The Task Force’s approach aimed to ensure that those most affected by AI's transformation of work could shape the State's response.
The Working Group used Policy Synth, a free and open source AI-based toolkit developed by Citizens Foundation and The GovLab, to synthesize the findings from research and engagements with private and public sector workers in the state. The Working Group used this approach to develop evidence-based policies while also enhancing democratic participation.
As a result of this process, which the Working Group undertook over eight weeks, New Jersey is implementing free AI skills training for all public servants and developing an AI-powered labor market monitoring system to help workers navigate career trends.
How it Worked
The Task Force used a two-pronged approach to develop its recommendations to State leadership: 1) using AI agents to conduct large-scale automated web research and 2) collecting direct input from New Jersey workers to understand the impact of AI on the state's workforce.
AI-Enabled Automated Research
To generate the first set of solutions, we used the Policy Synth tool kit to gather insights from existing research about the problem and possible solutions. We:
-
Generated a list of thousands of issues based on large-scale automated web research. In response to a written prompt Policy Synth’s AI agents searched thousands of online sources to generate a list of challenges related to the potential negative impacts of AI on New Jersey workers, the state’s economy and job market. AI allowed us to search and draw insights from a larger set of sources more rapidly than would have been possible through a traditional bench research process.
-
Prioritized among issues – Policy Synth’s AI agents scored and ranked the list of problems based on instructions we provided. For example, the agents prioritized problems based on criteria such as the problem’s anticipated economic impact, the number of individuals affected, and the urgency of addressing the issue within the next decade. This ranking helped us to narrow from a list of thousands of potential issues into a shorter list of 15 priority issues to focus on.
-
Generated a list of solutions – Next, we conducted large-scale, automated web research to identify solutions that responded to the 15 highest-ranked issues. Similar to the problem search phase, Policy Synth’s AI agents crawled websites, academic journals, white papers, and other online sources, then wrote up the ideas using a large language model (LLM).
-
Evolved solutions – Policy Synth then “evolved” the solutions using a genetic algorithm. The algorithm combined, added and removed various aspects of each solution to test whether those changes improved the quality of the solution. Through ten rounds of this mutation, the system transformed solution ideas into a set of hundreds of policy proposals outlining how each solution would work, expected benefits, and possible barriers to implementation.
-
The result? 1,451 policy proposals across 20 issue areas.
Engagement With Workers
To develop the second set of policy solutions we used the online engagement tool All Our Ideas to prioritize a set of problems to focus on. We:
-
Generated a list of problem statements – Using ChatGPT 4, Gemini, and Claude, we generated a list of 96 challenges when it comes to the impact of AI on the State’s economy and workforce. AI assisted in crafting concise, clear statements suitable for public engagement.
-
Prioritized among problems through public input – The Working Group asked workers to share their greatest hopes and concerns about the impact of GenAI on work in the Garden State using the online engagement tool All Our Ideas. Over three weeks in August, more than 2,200 private sector workers helped to create a rank-ordered list of the 96 problems. (Click here to read more about the engagement).
-
Developed solutions to priority issues – We then used Policy Synth to generate solutions to the top 20 issues identified through the public. Using the process described above, we generated solution ideas and then evolved the best solutions into policy proposals using a genetic algorithm.
-
The result? 1,101 policy proposals across 15 issue areas.
To narrow down the policy solutions, the Working Group identified eight priority issue areas that appeared in both problem searches:
- Economic Disparity
- Reduction in Overall Labor Demand
- AI-driven Occupational Shifts
- Worker Displacement
- Lack of Transparency and Accountability in AI Systems
- Impacts on Mental Health
- Impacts on Older Workers
- Privacy and Surveillance Challenges
We then considered the top five solutions for each of the eight priority problem areas, narrowing down the list of 2,500 policy proposals into a shortlist of 40 for consideration.
From Analysis to Action: Key Recommendations
The Task Force distilled the best and most actionable ideas from the top 40 policy proposals into four key recommendations for state leadership:
- Expand AI-integrated Skill Development: Enhance education and workforce training programs to incorporate AI skills development opportunities for all New Jersey residents
- Enhance NJ Career Navigator: Develop an AI-powered labor market monitoring and response system to help workers navigate changing career landscapes
- Expand Transition Support: Strengthen state workforce programs that support workers through job transitions
- Enable Small Business AI Adoption: Help Small Office/Home Office (SOHO) businesses leverage generative AI to drive growth
To arrive at these recommendations, policy experts on the Task Force chose ideas that were supported by evidence from a literature review of three dozen articles looking at the expanded impact of AI on skills demand and how to aid workers in navigating AI-driven career changes. To transform these ideas into actionable recommendations, the Task Force focused on expanding and scaling existing policies and programs that equipping workers and policymakers with the tools needed to tackle future challenges.
The Task Force also prioritized solutions that respond to workers’ concerns. For example, a survey of 5,000 public servants in New Jersey administered by the Working Group showed that three in four (73%) respondents want to learn more about how they can use AI in their work. As suggested in Recommendation #1, the state is rolling out free, comprehensive AI skills training to all public servants in New Jersey through InnovateUS. To take another example, results from the engagement with private sector workers showed that many had concerns that employers could implement AI systems in ways that displace workers. Recommendations #2 and #3 are improvements to existing programs designed to aid workers in anticipating and navigating career transitions – whether those transitions are AI-driven or due to larger changes in the job market.
To read the recommendations in full, click here.
The Task Force's innovative use of AI to develop policy and program recommendations about AI represents a new frontier in evidence-based policymaking.
The Power of Combining Technology and Human Expertise
What makes the Task Force’s approach particularly noteworthy is how it combined the computational power of AI with human expertise. While Policy Synth’s AI agents conducted the broad-scale research and initial analysis, subject matter experts on the AI Task Force reviewed outputs at each stage. While Policy Synth did not invent new solutions from scratch, its AI agents helped the Task Force to gather evidence of what works and similar solutions that have been implemented in other contexts. That evidence enabled us to design policy recommendations uniquely suited to New Jersey’s workforce needs, economic landscape, and policy environment.
The Task Force's innovative use of AI to develop policy and program recommendations about AI represents a new frontier in evidence-based policymaking. By leveraging technology to process vast amounts of information while maintaining human oversight and incorporating public input, New Jersey’s Task Force pioneered a model that other states might follow as they work to prepare their workforces for an AI-driven future.
Dane Gambrell is a Research Fellow at the GovLab supporting New Jersey’s AI Task Force. To read the Task Force’s full report, click here.
