Thread by @emilymbender

Looking forward to #chiir2022 this week and to presenting "Situating Search", with @chirag_shah
dl.acm.org/doi/10.1145/34…
1/

@chirag_shah In this #chiir2022 perspective paper we argue that using language model driven conversation agents (e.g. LaMDA) for search is flawed both technically and conceptually.
2/

Technical flaws include the fact that the language models aren't designed to perform "reasoning" (despite wild claims, such as Metzler et al 2021 referring to their "reasoning-like capabilities"). See also Bender & Koller 2020:
aclanthology.org/2020.acl-main.…

3/

The conceptual flaws have to do with the vision of how web search relates to human information seeking behavior: we need tools that help us surface and contextualize information sources, not systems that authoritatively provide 'answers'.
4/

In the paper, we explore different Information Seeking Strategies (ISS; Belkin et al 1993), and observe how a dialogue agent cuts off support for increasing information literacy and flexible exploration in most.
5/

We also explore how language model based dialogue agents can exacerbate problems such as the representational harms documented by Sweeney 2013 and Noble 2018:
6/

Finally, we present some thoughts about paths forward, including a call for transparency at many levels and desiderata for ideal search systems.
7/

So for those at #chiir2022, we're presenting in Viewpoints and perspectives 1 and (Mar 15, 7pm PST) and Viewpoints and perspectives 2.b (Mar 17, 8:30am PST). @chirag_shah and I hope to see many of you there!
/fin