Adoption Phasing and Timing
The integration of AI in legislative processes and institutional operations demands a phased approach. Initially, employing third-party applications and existing LLMs could provide useful augmentation, albeit necessitating quality and ethical oversight. Immediate actions could encompass enabling small-scale experimentation, fostering inter-office information sharing through transparent reporting, commissioning expert studies, and monitoring constituent interactions to gain insights into AI's practical utility.
In the medium term of six months to two years, the emphasis could shift towards investments in technical infrastructure to seamlessly embed AI in routine processes and to assemble multidisciplinary teams to harness AI in complex policy and oversight tasks, thereby augmenting the effectiveness of existing staff. Engaging with vendors to prototype customized AI applications tailored for legislative functions can accelerate AI adoption during this phase.
Looking further into the future, post two to five years of deliberate experimentation and integration, certain legislative bodies might be well-placed to revolutionize constituent engagement, lawmaking, and oversight using AI systems custom-built for their institutions. This phase could witness the development of in-house models trained on parliamentary data, leading to AI assistants adept in legislative processes and capable of drafting legislation in line with policy objectives. However, this requires substantial funding, multidisciplinary teams, and evolved oversight mechanisms to maintain human accountability.
In an April 2023 briefing, POPVOX Foundation shared the following recommendations with the US Congress. A review of subsequent steps taken in 2023 demonstrates that the institution independently took an approach that hewed closely to these suggestions:
Immediate (~3 months)
Facilitate experimentation; clarify tool incorporation guidelines
Promote information sharing (e.g., via House AI Working Group)
Acquire insights from other industries and legislatures
Track advocacy practices and constituent engagement for inauthentic campaigns
Short-term (3–6 months)
Conduct hearings in pertinent committees
Commission expert studies, engaging entities like GAO Innovation Lab
Draw lessons from private sector innovations and global instances
Encourage CRM vendors to integrate AI tools in offerings
Advocate for prototype development (e.g., through House Digital Services)
Medium-term (6 months–2 years)
Amplify investment in tech/IT modernization for integrated information flow and process automation
Formulate multi-disciplinary teams for complex policy, oversight tasks, and staff relief
Long-term (2–5 years)
Revolutionize constituent interaction, lawmaking, policy modeling, and oversight practices
