Elektrine
Log in Register
Paige Chat Timeline Gallery Friends Email Drive DNS Private DNS Domains VPN Kairo Nerve
Remote

Séverin Lemaignan

@severin@hci.social
mastodon 4.7.3
  • Open on hci.social

Social robots tamer & Senior scientist #SocialRobotics #HRI and AI @CSIC @IIIA

102 Followers
80 Following
19 Posts
Joined October 30, 2022
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
@jzb@hachyderm.io I fully agree: it is certainly not the way today, with the current ecosystem dominated by a handful of large US and Chinese companies. But LLMs, as a science and technology, are not intrinsically bound to these companies. Up to us to train and develop ethical solutions. And it is not a theoretical possibility: see #allenai #EuroLLM #Aperture
1
0
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
Until recently, this would have required being a solid coder. Not anymore. I deeply believe that LLMs will empower people, like nothing before, to _own_ the technology they depend on. I wish the open-source community would realise that. LLM might disrupt today's open-source workflows, and force a deep transformation. But LLMs might also become an extremely powerful vector to support and expand the _idea_ of libre software. Let invent the processes, practices, tools to make that happen! 9/9
1
8
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
Even though I've written code for 15+ years, with 100s of pre-LLM open-source repositories, I am not really interested in the code itself. In fact, I'm very happy for an LLM to do that for me. What I deeply care about, however, is to build excellent open-source tools, and make them available to everyone. Tools that are as good or better than proprietary ones. I want to contribute to a world where people own their tech stack, and do not depend on the whims of a few capitalist actors. 8/9
1
11
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
But the important point is elsewhere: I did not submit this PR because I really wanted to join the community of Inkscape developers. I submitted this PR because it wanted to address a pain point in the software. I did it (with the help of an LLM) and I thought it could be useful to others, so I tried (and failed) to share it. 7/9
1
13
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
It seems to me that the open-source community is torn between conflicting feelings atm: real, difficult technical challenges that might force an unpleasant re-thinking of how to 'do open-source'; the subjective feeling of being somewhat robbed of a hard-gained expertise and peer recognition by no-effort 'LLM kiddies'; but also the recognition that LLMs are extremely powerful tools, probably much better that the average OS programmer, and not likely to be simply waved away anyway. 3/9
1
17
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 7mo ago
Replying to
@delta I share the feeling about deltachat loosing its 'unique value proposition' (the transparent 'merging' of emails and chat). The technical argument 'we are still 100% using the email infrastructure' is true, but also kind of moot if I can not anymore (easily) use my email address as handle. Combined with the by-design lack of central contact discovery, how am I supposed to know if my contacts are on deltachat? 'In-person'-only discovery is not great UX imho
3
0
1
0
Open post
Séverin Lemaignan @severin@hci.social
· 6mo ago
Replying to
@EUCommission what does the 'research' side of the agreement entail? Will Australia join the Horizon Europe research framework?
1
1
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 6mo ago

Miriam presenting our work on generating robot faces procedurally for our #Emy social robot. @hri@hci.social #SocialMinds

hci.social

🌱 hci.social

1
0
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 16mo ago
Replying to
@haeckerfelix@mastodon.social congrats!! #twig us really what connects me to the community. I love it!! Thanks so much!🤗
2
0
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 26mo ago
Replying to
@chriswood@mastodon.design great write up! If you don't know him yet, I encourage you to check the work of @davidrevoy@framapiaf.org -- he is a very respected digital artist in the Linux community, and does regularly extensive testing of free software/hardware setup on Linux, like https://www.davidrevoy.com/article1030/debian-12-kde-plasma-2024-install-guide
Debian 12 KDE Plasma: The right GNU/Linux distribution for professional digital painting in 2024. Reasons and complete installation guide.
David Revoy

Debian 12 KDE Plasma: The right GNU/Linux distribution for professional digital painting in 2024. Reasons and complete installation guide.

4
2
1
0
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
Accepting a PR from a new contributor implicitly means accepting the contributor her/himself into the project: a technical contribution in exchange of social recognition. But this only make sense as long as the contributor has actually contributed a significant effort. In my case, I did not code anything. Were defining a problem, supervising a LLM, testing the results, participating to the follow-up discussions, sufficient contributions? maybe not. @doctormo@floss.social you might have a different take? 6/9
0
14
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
There was a good technical reason for it, but at this stage, the back-and-forth and permanent need for justification was draining, and I decided to eventually withdraw my PR. In that case, I think the negative feelings were as much about the objective risks (who would maintain that code, etc), as they were about the subjective feelings that accepting this kind of contribution would open a pandora box of (potentially useful) LLM features that do not deeply 'belong' to the project ethos. 5/9
0
15
0
1
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
One example: I submitted a PR to Inkscape https://gitlab.com/inkscape/inkscape/-/merge_requests/7862 to improve a specific algorithm. The code was entirely written by Claude.My role & contribution were purely about defining the problem, supervising the coding process and testing the result. The code was initially well received (as the feature was useful), but the feeling changed when @doctormo@floss.social one of the lead maintainers, found that the LLM code duplicated a specific colour conversion algorithm already implemented elsewhere. 4/9
gitlab.com

Add custom color palette support to bitmap tracing with novel color quantization algorithm (!7862) · Merge requests · Inkscape / inkscape · GitLab

0
16
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 2mo ago
Replying to
The critique is valid and focuses on the *coding* side of open-source. One can argue whether 'hand-coding' will still be relevant in a few years, or whether it will become a niche craft, like coding directly in assembly today. But there's no denying that the slow-paced community-building process on which open-source thrive -- and where individual contributors find meaning and social reward -- is hardly compatible with the primarily individualistic & fast-paced nature of LLM code churning. 2/9
0
18
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 6mo ago
Replying to
@delta looks like an impressive release! Audio/video calls based on WebRTC 🤩 Congrats to the devs! imho, the last remaing hurdle for much greater reach for deltachat is contact discovery. Any plans/ideas about that?
0
2
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 6mo ago

At the #HRI2026 conference, I presented our thoughts on the current #HumanoidHype: yes, that's a lot of undeniably cool machines, doing backflips and complicated dances, but well... it feels that the US and Chinese companies leading the show haven't fully realized/understood what it means to create robots to work alongside humans.

More here: https://www.linkedin.com/posts/skadge_hri2026-humanoidhype-responsiblerobotics-share-7442217711637749761-BJzN?utm_source=share&utm_medium=member_desktop&rcm=ACoAAACXCd8B5Q2e7tT7qrUwwsR-eymaaKR0ceg

You can read the full paper here: https://lnkd.in/deSmD7dK

hci.social

🌱 hci.social

0
0
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 6mo ago

@BenjaminRosman keynote at #HRI2026 : strengthening the ML community in Africa. Now running the largest summer school in AI in the world. Inspiring! #SustainableHRI

hci.social

🌱 hci.social

0
0
0
0
Open post
Séverin Lemaignan @severin@hci.social
· 4mo ago
Replying to
@0xf0@camp.smolnet.org ...if only it was any good... 15 years that I _have_ to use it (...main/only option on Linux), and I still rage-quit it after 10min everytime I open it. I can see why one might want to support it for political reason (and that's a good enough reason), but man... The UX was already bad 15 years ago, and it hasn't changed a bit since then.
0
0
0
0
Back
313k7r1n3
Elektrine

Tor hidden service

elekhj7afj4qnrr4yd3bkzslsyo5jgfxw3orgjkhlcxifueodybyiiad.onion

I2P eepsite

j6b6cyk6gjmepjih7jjadxgxvvf3lzzujljuu2v4biemzpg3naya.b32.i2p

Platform

  • Email
  • Chat
  • Timeline
  • VPN
  • DNS

Company

  • About
  • Contact
  • FAQ
  • Lite (no JS)

Legal

  • Terms of Service
  • Privacy Policy
  • Transparency Report
  • Report Abuse
  • Warrant Canary
  • VPN Policy

Support

  • support@elektrine.com
  • Report Security Issue
Mail client setup IMAP mail.elektrine.com:993 POP3 mail.elektrine.com:995 SMTP mail.elektrine.com:465
© 2026 Elektrine. All rights reserved. Server: 08:27:19 UTC