We are no longer active here, but you are welcome to follow us on our other channels. Find out more in the pinned tweet. ⬇
Berlin
Joined November 2015
- Tweets4.2K
- Following122
- Followers17.1K
- Likes2.4K
AlgorithmWatch retweeted
For anyone able to work in German and Germany, @algorithmwatch has two cool jobs at the intersection of comms and tech-policy. Please do pass on if you know anyone who might fit.
algorithmwatch.org/de/job-se…
algorithmwatch.org/de/job-le…
AlgorithmWatch retweeted
What happens when you ask an AI chatbot questions about upcoming elections?
According to @aiforensics_org and @algorithmwatch, Microsoft's Bing AI chatbot gave incorrect information 1/3 of the time when prompted to answer questions about candidates and voting procedures in German and Swiss elections.
Via @washingtonpost & @WillOremus:
wapo.st/48prJRr
AlgorithmWatch retweeted
ChatGPT, Bard or Grok –
AI-controlled #chatbots may answer questions quickly, but not always accurately.
They can distort facts as they are trained to always use the next most likely word in a sentence.
@AlgorithmWatch explains why this matters in #DWdigital's latest video⤵️
AlgorithmWatch retweeted
#LandeckerRecap2023
This year, our partner @AlgorithmWatch launched “Auditing Algorithms for Systemic Risks” to evaluate #algorithms & their impact. The project aims to dismantle the “black box” of algorithms and increase the transparency and accountability of platforms.
AlgorithmWatch retweeted
As part of the #ShapingAI closing conference, HIIG, @ZeMKI_Bremen & @algorithmwatch are organising a public panel discussion at @mfk_berlin. @NoortjeMarres discusses the intertwining of AI & society with @BJMcP, @markscott82 & @spielkamp.
Registration: hiig.de/en/events/evening-pa…
AlgorithmWatch retweeted
New research from a pair of European nonprofits finds that Microsoft’s Bing AI chatbot, recently rebranded as Microsoft Copilot, gave inaccurate answers to 1 out of every 3 basic questions about candidates, polls and scandals. wapo.st/3Tnu7DN
AlgorithmWatch retweeted
Research from @algorithmwatch and @aiforensics_org published this morning showed that Copilot produced factual errors 1/3 of the time when asked about elections in Switzerland and Germany earlier this year.
Including candidate lists, polling data, voting locations and times
2/7
AlgorithmWatch retweeted
🚩Our research w @algorithmwatch revealed: Election misinformation by AI-driven chatbots such as Microsoft Co-pilot (formerly Bing Chat) is systemic and not just isolated cases of "hallucinations".
@WIRED & @daithaigilbert covered it:
wired.com/story/microsoft-ai…
AlgorithmWatch retweeted
On 8 Dec, after over 36 hours of closed-door negotiations, the EU reached an agreement on the #AIAct. Key details are yet to be determined, but #CSOs fear the Act will fall short of effectively protecting people from harmful effects of AI.
Learn more: ecnl.org/news/eu-reaches-agr…
AlgorithmWatch retweeted
🚨 EU: Do not trade away our rights in the #AIAct!
Civil society advocates and academics are calling on the @Europarl_EN, @EUCouncil & @EU_Commission to unequivocally reject attempts to legalise dangerous, discriminatory police AI
🧵 #DontTradeOurRights
edri.org/wp-content/uploads/…
ALT Open letter to EU AI Act negotiators: Do not trade away our rights! As the supposedly final negotiation on the EU’s landmark Artificial Intelligence (AI) Act drags on, the undersigned civil society coalition calls on lawmakers to unequivocally reject attempts to legalise dangerous and discriminatory police AI. EU lawmakers are being pressured by governments and security industry lobbyists to trade away our fundamental rights and freedoms. The Parliament’s position on prohibitions is backed by the UN High Commissioner for Human Rights, the European Data Protection Supervisor and Board, networks of national human rights institutions and equality bodies, academics and other experts.
AlgorithmWatch retweeted
This episode we looked at AI & elections across 2 continents. We met fact checkers in 🇰🇪 @FumbuaKE & @EstherKazungu , researchers who advise 🇪🇺politicians about deepfakes, & Bing chat investigators @algorithmwatch.
For TV w/ SHIFT @dwnews. Online here: youtu.be/0QlScDJtqWw?si=PsUZ…
AlgorithmWatch retweeted
Negotiations on the #AIAct will resume tomorrow morning.
Now more than ever, the @Europarl_EN @brandobenifei @IoanDragosT must show us they will #ProtectNotSurveil every person by dangerous AI systems.
What’s at risk in🧵
AlgorithmWatch retweeted
1/2 🚨The EU is trying to expand #EURODAC, a database to facilitate the registration and deportation of asylum seekers
The expansion includes:
🔎Use of #FacialRecognition
🧒🏽Including children as young as 6
🚔More access by police
Civil society says: #StopEURODAC
AlgorithmWatch retweeted
Friederike Rohde im @rbb24: Beim #Energie-Sparen hilft es, #KI-Modelle schlank zu halten, denn große Rechenmodelle wie GPT3 mit vielen Parametern sind energieintensiv. Und: Die #KIVerordnung sollte Unternehmen verpflichten, ökolog. Kennzahlen zu erheben. rbb-online.de/abendschau/7pl…
ALT Live-Interview von Friederike Rohnde im Studo der rbb-24-Abendschau
In der Klimaminute der @rbb24 Abendschau spreche ich heute über #KI & #Nachhaltigkeit und die Klimafolgen von #MachineLearning.🧐 Schön, dass unsere Arbeit im Projekt #sustAIn @ioew_de @algorithmwatch zur öffentlichen Debatte beiträgt. Wir brauchen mehr Transparenz und Messdaten!
AlgorithmWatch retweeted
1/3🚨As we head into final negotiations on the #AIAct next week, civil society is calling on @EUCouncil to change course and protect our #HumanRights by effectively regulating the use of AI systems by law enforcement.
Read our statement⤵️
edri.org/our-work/civil-soci…
AlgorithmWatch retweeted
The #AIAct is at risk because Council and Parliament are unable to reach a common position on #GPAI & Foundation Models. None of the current "compromises" ensures adequate protection of our fundamental rights. Our Op-ed highlights crucial weaknesses: algorithmwatch.org/en/op-ed-…
AlgorithmWatch retweeted
Matthias Spielkamp sagt, #KI solle „demokratiefördernd und gemeinwohlorientiert“ sein. Für die Regulierung müsse ein Rechtsrahmen geschaffen werden, so der KI-Experte.
AlgorithmWatch retweeted
Algorithmen können Menschen diskriminieren - Doch was kann man dagegen tun?
Unsere Podcast-Gästin Pia Sombetzki von @algorithmwatch hat mit uns über algorithmische Diskriminierung gesprochen und was Betroffene tun können.
Viel Vergnügen!
netzpiloten.de/wie-algorithm…
AlgorithmWatch retweeted
Who sees which ad? Discriminatory ads continue to run rampant on Facebook. Meta has deployed a new AI system to fix its algorithmic bias problem for US housing ads netzpolitik.org/2023/platfor…
AlgorithmWatch retweeted
Great contribution by @algorithmwatch’s @AnMollen on workplace algorithms and workers’ participation in algorithms design at #EUSocialForum. Risk mitigation is not the whole solution - workers need to participate in the design of software used for workers management.
