Conversational Search and Analysis of Collections of Letters and Comments


Today, public-sector personnel can, increasingly, utilize conversational search engines over large collections of letters and comments sent to elected representatives or in response to regulatory rulemaking processes. As the two publications shared below indicate, delivering these entirely realizable technologies to the public sector would greatly benefit democracy.

Brainstorming, in the not-too-distant future, conversational AI agents could run scripts and interact with conversational search engines, on behalf of human personnel, to accelerate repetitive procedures involved with producing reports or dashboards.

That is, beyond engaging in multi-step man-machine dialogue about collections of letters and comments, public-sector personnel could dispatch conversational AI agents which would interact on their behalf, in a procedural manner, with other conversational AI systems, e.g., conversational search engines, about large collections of letters and comments while producing valuable reports and dashboards.

Interestingly, generated reports and dashboards could be open and transparent, public-facing and Web-based, so that citizens could also explore them.

Below are two publications - with selected quotes - about how AI could be of use for empowering public-sector personnel to process bulk letters and comments sent to elected representatives or in response to regulatory rulemaking processes.

AI Could Shore Up Democracy – Here’s One Way (link)

"Consider individual letters to a representative, or comments as part of a regulatory rulemaking process. In both cases, we the people are telling the government what we think and want.

"For more than half a century, agencies have been using human power to read through all the comments received, and to generate summaries and responses of their major themes.

"In the absence of that ability to extract distinctive comments, lawmakers and regulators have no choice but to prioritize on other factors. If there is nothing better, ‘who donated the most to our campaign’ or ‘which company employs the most of my former staffers’ become reasonable metrics for prioritizing public comments. AI can help elected representatives do much better.

“If Americans want AI to help revitalize the country’s ailing democracy, they need to think about how to align the incentives of elected leaders with those of individuals.”

Implementing Federal-wide Comment Analysis Tools (link)

"The federal government publishes tens of thousands of documents each year in the Federal Register, with over 800,000 total documents since 1994, which garner millions of submissions from the public (comments and other matter presented).

"Agencies have a legal obligation to consider all relevant submissions and response to those which, significantly, would require a change to the proposed rule. To discern relevance, significance, and disposition, human review is needed.

"The capacity for human review often can’t meet the demand for high-volume comment events. Initial screening and classification allow regulator officials to focus on relevant submissions and response to groups of significant comments address the same topic. Some agencies perform independent, tailored analyses to assist with this initial screening.

“The CDO Council recognized an opportunity to leverage recently advanced Natural Language Processing (NLP), which would be more efficient than these independent analyses. A generalizable toolset could provide effective comment grouping with less upfront effort, and this toolset could be shared and reused by rule makers across government to aid and expedite their comment analysis.”

Maybe, but Openai doesn’t allow such usage.

@jochenschultz, from a development perspective, this matter would seemingly pertain to the weighting or ranking of letters and comments in collections.

One approach would be to utilize a uniform distribution of weights across individuals’ letters and comments, to weigh the letters and comments equally. Another approach might involve prioritizing letters and comments from individuals in representatives’ districts over those from individuals in other districts. Perhaps there could be configuration settings which could be toggled so that end-users could access reports and dashboards generated from multiple views of the same input data.

I’m also visualizing weighting or ranking the embedding vectors from collections of input letters and comments, accumulating weights for these vectors as they occur in multiple documents, coalescing these vectors, doing other mathematics upon these sets of vectors, to subsequently generate summarizing reports or dashboards.

These are but preliminary ideas. I am eager to hear from you and others with respect to more ideas about uses of AI in processing the letters and comments sent to elected representatives or in response to regulatory rulemaking procedures.

Under “Usage policies” you will find:

  • Political campaigning or lobbying, by:
    • Generating high volumes of campaign materials
    • Generating campaign materials personalized to or targeted at specific demographics
    • Building conversational or interactive systems such as chatbots that provide information about campaigns or engage in political advocacy or lobbying
    • Building products for political campaigning or lobbying purposes

What do you think? Could it potentially have an impact on politics?

Aha, thank you, now I understand.

As I look at that list, the use cases of processing and analyzing collections of letters and comments sent to elected representatives or in response to regulatory rulemaking procedures appear to be entirely distinct from those prohibited scenarios pertaining to dystopian campaigning and lobbying.

Also, that list of prohibited scenarios doesn’t appear to be intending to prohibit all applications which potentially have “an impact on politics”.

As I understand, the prohibited scenarios appear to be addressing important concerns resembling those raised here: How AI Could Take Over Elections and Undermine Democracy.

I mean I personally don’t have any problems with that. I am more like trying to avoid more posts about “my account got blocked, what can I do to reactivate it”.
Not that I got any other informations than that in the Usage policies.

But I’d ask if the use case is ok…

bottom right corner there is a chat window

Yes, that makes sense. OpenAI developer relations and support staff are always available for clarification.

Assuming that the use cases under discussion are not prohibited by OpenAI’s policies, those teams interested in enhancing elected representatives’ capacities to process and analyze incoming letters and comments from their constituencies and/or, similarly, in enhancing regulatory agencies’ capacities to process and analyze incoming letters and comments could visualize, imagine, and model the tasks and workloads of those personnel who work with these vast collections of documents, sometimes in time-critical scenarios.

Beyond the information available here, interested teams could also search for other existing market analyses - or perform new such analyses - to ascertain more specific needs of these personnel with which to develop quality tools, products, and services.