Harish Tayyar Madabushi
@harish
Followers
2K
Following
2K
Statuses
2K
Lecturer (~Assistant Professor) in Artificial Intelligence. Work on Deep Learning for #NLProc and Deep Contextual Meaning Representations
Bath, England
Joined December 2008
Thank you Yann LeCun @ylecun for reading and sharing our #ACL2024NLP work! 🔖Merely scaling up LLMs is *not* enough! Summary Paper Coauthors: Sheng Lu @IBigoulaeva
@rachneet4 @IGurevych
#NLProc
Sometimes, the obvious must be studied so it can be asserted with full confidence: - LLMs can not answer questions whose answers are not in their training set in some form, - they can not solve problems they haven't been trained on, - they can not acquire new skills our knowledge without lots of human help, - they can not invent new things. Now, LLMs are merely a subset of AI techniques. Merely scaling up LLMs will *not* lead systems with these capabilities. There is little doubt AI systems will have these capabilities in the future. But until we have small prototypes of that, or at least some vague blueprint, bloviating about AI existential risk is like debating the sex of angels (or, as I've pointed out before, worrying about turbojet safety in 1920).
1
6
23
RT @davmoltisanti: 📯Today we release HD-EPIC📯 A new highly detailed egocentric video dataset. Check out Dima's thread below and visit https…
0
1
0
Happening at 2pm! Join us at the COLING 2025 tutorial: Bridging Linguistic Theory and AI: Usage-Based Learning in Humans and Machines with James Pustejovsky @jamespusto Nikhil Krishnaswamy and Claire Bonial #coling2025 #nlproc #tutorial
1
5
15
Excited to be at @coling2025! Looking forward to meeting collaborators in person and sharing our work. Here @melissatorgbi and I from @BathNLP.
0
2
26
RT @josephimperial_: 📣 Calling all domain experts and researchers working on interdisciplinary areas x #AI #NLP #LLMs Do you think AI can…
0
6
0
@tricalt Prior work seemed to indicate that LLMs have "emergent abilities". This implied an existential threat!! 📢 With evidence from over 1000 experiments, we demonstrated that LLMs do NOT pose an existential threat. More: Video:
Dr Harish Tayyar Madabushi @harish from the Department of Computer Science describes the pros, cons and limitations of large language models (LLMs), such as ChatGPT. 👇 @UniofBathSci #AI #ChatGPT #LLMs
0
0
1
RT @josephimperial_: Domain experts use standards (documented rules) to create text content that aligns with their varying levels of requir…
0
2
0
RT @josephimperial_: I'll be presenting (oral) one example of this work focusing on standards in education and language proficiency @ #EMNL…
0
2
0
RT @BathNLP: Our lab will be presenting 3 long papers at #EMNLP2024 in Miami, Florida! 🥳🌴 Topics cover controllable NLG w/ standards, benc…
0
3
0
RT @yoavgo: Wanna get a birds-eye view of the topics and trends in #EMNLP2024? check out the link in @UrikaUri 's thread, presenting navig…
0
2
0
1 in 6 vapes confiscated by schools have Spice in them!! ðŸšðŸš Work by researchers @UniofBath including @TomSFHaines raised in Parliament.
Local MP Wera Hobhouse raises Bath research in Parliament which found that the synthetic street drug Spice was being sold in vapes and being used by children.
0
1
9
@Sciprios @GHRNewsUK @trkjeldsen @CaerBath I found this work really interesting @Sciprios 🙌 Identifying flood risk 🌊 and ensuring the safety of maintenance teams ... using machine learning🤖!
0
0
0
RT @peterjansen_ai: Can language models perform end-to-end scientific discovery? In our NeurIPS Spotlight paper, we show: very rarely. Our…
0
105
0
