AIDA Research Group - UGent (@aidaugent) 's Twitter Profile
AIDA Research Group - UGent

@aidaugent

Official X account of the Artificial Intelligence and Data Analytics (AIDA) Group at Ghent University led by Prof. @TijlDeBie

ID: 1776226185678028800

linkhttps://aida.ugent.be/ calendar_today05-04-2024 12:31:58

31 Tweet

30 Followers

43 Following

Tijl De Bie (@tijldebie) 's Twitter Profile Photo

#LLMs are superhuman at persuasion. In a new AIDA Research Group - UGent-preprint, we survey recent research on this topic, identifying the factors affecting persuasiveness. We also discuss the profound ethical and societal challenges this poses. arxiv.org/pdf/2411.06837 European Research Council (ERC) UGent

#LLMs are superhuman at persuasion.
In a new <a href="/aidaugent/">AIDA Research Group - UGent</a>-preprint, we survey recent research on this topic, identifying the factors affecting persuasiveness.
We also discuss the profound ethical and societal challenges this poses.
arxiv.org/pdf/2411.06837
<a href="/ERC_Research/">European Research Council (ERC)</a> <a href="/UGent/">UGent</a>
Tijl De Bie (@tijldebie) 's Twitter Profile Photo

โ“Are #LLMs ideological? The release of DeepSeek R1 has revived this important question. โžก๏ธ We substantially extended our earlier study that confirms they are, now spanning 19 LLMs from across ๐Ÿ‡บ๐Ÿ‡ธ๐Ÿ‡จ๐Ÿ‡ณ๐Ÿ‡ท๐Ÿ‡บ๐Ÿ‡ซ๐Ÿ‡ท๐Ÿ‡ฎ๐Ÿ‡ฑ๐Ÿ‡ธ๐Ÿ‡ฆ๐Ÿ‡ฆ๐Ÿ‡ช๐Ÿ‡ฉ๐Ÿ‡ช & prompted in the 6 United Nations-languages, and including new analyses.

โ“Are #LLMs ideological?
The release of <a href="/deepseek_ai/">DeepSeek</a> R1 has revived this important question.

โžก๏ธ We substantially extended our earlier study that confirms they are, now spanning 19 LLMs from across ๐Ÿ‡บ๐Ÿ‡ธ๐Ÿ‡จ๐Ÿ‡ณ๐Ÿ‡ท๐Ÿ‡บ๐Ÿ‡ซ๐Ÿ‡ท๐Ÿ‡ฎ๐Ÿ‡ฑ๐Ÿ‡ธ๐Ÿ‡ฆ๐Ÿ‡ฆ๐Ÿ‡ช๐Ÿ‡ฉ๐Ÿ‡ช &amp; prompted in the 6 <a href="/UN/">United Nations</a>-languages, and including new analyses.
Anthropic (@anthropicai) 's Twitter Profile Photo

New Anthropic research: Do reasoning models accurately verbalize their reasoning? Our new paper shows they don't. This casts doubt on whether monitoring chains-of-thought (CoT) will be enough to reliably catch safety issues.

New Anthropic research: Do reasoning models accurately verbalize their reasoning?

Our new paper shows they don't.

This casts doubt on whether monitoring chains-of-thought (CoT) will be enough to reliably catch safety issues.
ECML PKDD (@ecmlpkdd) 's Twitter Profile Photo

๐Ÿ“ฃ Attention passengers! This is the flight to the biggest European machine learning conference. โœˆ๏ธ ๐——๐—ฒ๐—บ๐—ผ ๐—ง๐—ฟ๐—ฎ๐—ฐ๐—ธ ๐——๐—ฒ๐—ฎ๐—ฑ๐—น๐—ถ๐—ป๐—ฒ: ๐— ๐—ฎ๐˜† ๐Ÿฎ๐—ป๐—ฑ + info: ecmlpkdd.org/2025/submissioโ€ฆ #ECMLPKDD2025

๐Ÿ“ฃ Attention passengers! This is the flight to the biggest European machine learning conference. โœˆ๏ธ ๐——๐—ฒ๐—บ๐—ผ ๐—ง๐—ฟ๐—ฎ๐—ฐ๐—ธ ๐——๐—ฒ๐—ฎ๐—ฑ๐—น๐—ถ๐—ป๐—ฒ: ๐— ๐—ฎ๐˜† ๐Ÿฎ๐—ป๐—ฑ + info: ecmlpkdd.org/2025/submissioโ€ฆ #ECMLPKDD2025
Google AI (@googleai) 's Twitter Profile Photo

We just announced Geospatial Reasoning, a new research effort that will use generative AI (#Gemini) to help unlock novel insights about our world. ๐ŸŒ Imagine asking complex Qs about our planet & getting detailed answers w/ plans & visuals! goo.gle/3XOO5sF

We just announced Geospatial Reasoning, a new research effort that will use generative AI (#Gemini) to help unlock novel insights about our world. ๐ŸŒ

Imagine asking complex Qs about our planet &amp; getting detailed answers w/ plans &amp; visuals! goo.gle/3XOO5sF
AIDA Research Group - UGent (@aidaugent) 's Twitter Profile Photo

๐Ÿง  Can we still trust science? Peer review, self-correction, disinterested truth-seekingโ€”once solid foundations, now under pressure. Join us for a critical look at scienceโ€™s gatekeeping mechanismsโ€”and why generative AI might worsen the cracks. ๐ŸŽ™ How to trust science? Prof.

AIDA Research Group - UGent (@aidaugent) 's Twitter Profile Photo

Disinformation & Artificial Intelligence - a toxic mix? Next week Thursday evening, April 24, Tijl De Bie will participate in a debate on this burning topical topic in Ghent. About the thin line between #information and #manipulation, between #regulation and #censorship. About

Disinformation &amp; Artificial Intelligence - a toxic mix?

Next week Thursday evening, April 24, <a href="/TijlDeBie/">Tijl De Bie</a>  will participate in a debate on this burning topical topic in Ghent.

About the thin line between #information and #manipulation, between #regulation and #censorship.

About
Tijl De Bie (@tijldebie) 's Twitter Profile Photo

How to find #humanity in an #AI-flooded world? That's the question we'll discuss in the #EcceMachina symposium, on Wed 14 May 14:00-18:00 at UGent AIDA Research Group - UGent (Auditorium Vandenhove). Open & free to all, but registration is mandatory. (Registration links in next message:)

How to find #humanity in an #AI-flooded world?

That's the question we'll discuss in the #EcceMachina symposium, on Wed 14 May 14:00-18:00 at <a href="/ugent/">UGent</a> <a href="/aidaugent/">AIDA Research Group - UGent</a> (Auditorium Vandenhove).

Open &amp; free to all, but registration is mandatory. (Registration links in next message:)
AIDA Research Group - UGent (@aidaugent) 's Twitter Profile Photo

The list of workshops organised at ECML PKDD is online ๐Ÿฅณ Our research group AIDA Research Group - UGent is co-organising three of them: - The first workshop on Human-Centered Data Mining (HuMine) - The international workshop on AI for Human Resources and Public Employment Services (AI4WORK)

The list of workshops organised at <a href="/ECMLPKDD/">ECML PKDD</a>  is online ๐Ÿฅณ 

Our research group <a href="/aidaugent/">AIDA Research Group - UGent</a>  is co-organising three of them:
- The first workshop on Human-Centered Data Mining (HuMine)
- The international workshop on AI for Human Resources and Public Employment Services (AI4WORK)
Tijl De Bie (@tijldebie) 's Twitter Profile Photo

The AIDA Research Group - UGent research group at UGent in Belgium has multiple fully-funded PhD-scholarships available, on the European Research Council (ERC) Advanced Grant VIGILIA (โ€œVIrtual GuardIan angeLs for the post-truth Information Ageโ€). If you are interested in how generative AI, knowledge graphs, and

The <a href="/AIDAugent/">AIDA Research Group - UGent</a> research group at <a href="/UGent/">UGent</a> in Belgium has multiple fully-funded PhD-scholarships available, on the <a href="/ERC_Research/">European Research Council (ERC)</a> Advanced Grant VIGILIA (โ€œVIrtual GuardIan angeLs for the post-truth Information Ageโ€).

If you are interested in how generative AI, knowledge graphs, and
ECML PKDD (@ecmlpkdd) 's Twitter Profile Photo

#ECMLPKDD2025: Dear passengers, ๐˜„๐—ฒ ๐˜„๐—ถ๐—น๐—น ๐—ฏ๐—ฒ๐—ด๐—ถ๐—ป ๐˜๐—ต๐—ฒ ๐—ฑ๐—ฒ๐˜€๐—ฐ๐—ฒ๐—ป๐˜ ๐—ผ๐—ณ ๐—ผ๐˜‚๐—ฟ ๐——๐—ฒ๐—บ๐—ผ ๐—ง๐—ฟ๐—ฎ๐—ฐ๐—ธ ๐—ณ๐—น๐—ถ๐—ด๐—ต๐˜ with destination Porto in a few moments. ๐—˜๐˜…๐—ฝ๐—ฒ๐—ฐ๐˜๐—ฒ๐—ฑ ๐˜๐—ถ๐—บ๐—ฒ ๐—ผ๐—ณ ๐—ฎ๐—ฟ๐—ฟ๐—ถ๐˜ƒ๐—ฎ๐—น: ๐Ÿฎ๐Ÿฏ:๐Ÿฑ๐Ÿต ๐—”๐—ผ๐—˜, ๐— ๐—ฎ๐˜† ๐Ÿฎ๐—ป๐—ฑ. + info: lnkd.in/dheXAejB

AIDA Research Group - UGent (@aidaugent) 's Twitter Profile Photo

๐ŸŒ The 21st-century job market is changing faster than ever. AI and automation are reshaping entire industries, making reskilling, fair hiring, and data-driven workforce planningurgent priorities. ๐Ÿ“ข Join us at the AI4WORK Workshop (ECML PKDD 2025) to explore how AI, LLMs, and