NEW study recruitment📢 Join to help us design the next generation #Copilot experience!! #ai #hci #design

We’re hiring! Come work on amazing applications of LLMs to spreadsheet code and data, with a welcoming and impactful team at Microsoft Research in Cambridge UK linkedin.com/pulse/internsh… #excel

We're #hiring another #HCI #research #intern to explore #collaboration in #hybrid #blendedreality, incl. using LLMs. Join us at Microsoft Research UK this Summer! careers.microsoft.com/us/en/job/1518…

The VL/HCC 2023 deadline is coming up next month and we're excited to see your great work! 🗓️ Abstracts due April 21, 2023 🗓️ Papers due April 28, 2023 See the full call for papers here: conf.researchr.org/track/vlhcc-20…

No, it’s not an entirely new field unexplored. The field has existed for more than 40 years. Speaking of which, the flagship conference is happening just next week in Hamburg. ACM SIGCHI #CHI2023

<Summer Research Internship at KIXLAB> I'm looking for undergrad research interns to join my research group (kixlab.org) this summer. Many projects around designing and building novel human-AI interaction for various domains. juhokim.com/2023-summer-in…

2 papers just accepted at VLHCC 2023! (1) FxD: a functional debugger for dysfunctional spreadsheets and (2) ColDeco: An End User Spreadsheet Inspection Tool for AI-Generated Code (work led by our intern @kasra_ferdowsi)

more information here!: microsoft.com/en-us/research…

Also happy to announce:

MSR blog post about our two publications at VL/HCC 2023:

Looking forward to presenting at SPLASH 2023 (Onward! Essays) next week! 🎉 The essay, “Will code remain a relevant user interface for end-user programming with generative AI?”, probes the possibility that generative AI will make coding obsolete for non-expert programmers.

You need to audit AI output differently from human output because it makes different kinds of mistakes. E.g., AI is unlikely to be adversarial but more likely to hallucinate. Co-audit tools (arxiv.org/pdf/2310.01297…) help with evaluating AI output. #Slide18 aka.ms/nfw2023pdf#pag…

hey that’s us

What if we could directly edit the *assumptions* that LLMs make when solving end-to-end coding tasks? Our new #UIST2024 paper shows how decomposing a task into editable and progressively disclosed components enhances control and verification. Pre-print: arxiv.org/abs/2407.02651

Our CHIWORK paper on generative AI data analysis workflows has been highlighted on the MSR Research Focus blog along with other great work at MSR. Check it out here: microsoft.com/en-us/research…

Excited to present our paper at #UIST2024 tomorrow in the "AI as Copilot" session (4:10 PM) I'll be discussing how breaking down data analysis tasks into *editable*, *progressively revealed* steps can enhance control over AI and improve verification. 📜 dl.acm.org/doi/10.1145/36…