diogocortiz commented 1 year ago

Session description

In this session, we will discuss the possible impacts of generative AI on the Web. As it is an emerging topic and its consequences still need to be better understood, we will focus on technical and governance challenges in three main dimensions: data collection on the Web, Integration of LLMs in a search engine; production and publication of synthetic contents on the Web.

To facilitate the discussion, we will start from three main questions, which can be discussed in different groups depending on the number of participants:

What are the limits of scraping web data to train Generative AI and what technical measures should be implemented to ensure privacy, prevent copyright infringement, and effectively manage content creator consent?
What are the potential impacts of incorporating LLMs (Large Language Models) as chatbot interfaces into search engines? How would this change the web traffic and digital economy once the chatbot could respond directly to users instead of providing links?
How could Web technologies help detect AI-generated content posted on the Web, restrain the dissemination of sensitive content, and provide accountability?__

Session goal

The session aims to introduce the theme and discuss the possible consequences of Generative AI for the Web. We also want to stimulate a debate with people from different backgrounds on how Web technologies should respond to those changes. We also plan to elaborate and consolidate the result of the session to guide further discussion on the topic.

Additional session chairs (Optional)

@reinaldoferraz, Vagner Diniz

IRC channel (Optional)

GenerativeAI

Who can attend

Anyone may attend (Default)

Session duration

60 minutes (Default)

Other sessions where we should avoid scheduling conflicts (Optional)

No response

Estimated number of in-person attendees

Don't know (Default)

Instructions for meeting planners (Optional)

No response

Agenda, minutes, slides, etc. (Optional)

llemeurfr commented 1 year ago

I'm interested in the first aspect (what technical measures should be implemented to ensure privacy, prevent copyright infringement, and effectively manage content creator consent?).

I will represent the TDM Reservation Protocol CG (https://www.w3.org/community/tdmrep/), if the session if not too late on Wednesday.

jaykishigami commented 1 year ago

This is a very important and urgent topic, I was thinking about Breakout too, but I will join here.

I think there is a possibility that in the future, Generative AI will replace the role of search in the broader context of the web and the platform for providing information. In relation to the web, I think there could be a concept of a "Prompt Web". That is, currently there is an emphasis on prompting AI to provide good results through a variety of creative means, but this cannot be provided by the standardized web. In other words, prompts for AI to provide good results are currently regarded as important. I would like to talk from such a viewpoint.

npdoty commented 1 year ago

My only concern is whether that first item (technical mechanisms, for indicating opt-out/consent/preferences about training or data-mining) is sufficiently broad and interesting that we should have a separate dedicated session just for that.

AramZS commented 1 year ago

@llemeurfr I'm very interested in TDM, I think it might even be useful to have an entire separate session around that topic and the issues associated with it. Is the TDM group meeting as well? I don't think I saw them on the TPAC schedule.