Laut AGB will Zoom Kundeninhalte und Nutzungsdaten für KI-Lernzwecke verwenden. 😕 Das kann man versuchen, aber ich könnte Zoom als Anwalt nicht mehr nutzen und muss eigentlich jedem davon abraten Zoom im Rahmen von Kunden, Mandanten- oder Patientenkommunikation einzusetzen. Außer mit einer ausdrücklichen Einwilligung, die diese Nutzung nachweisbar und verständlich erklärt (was in der Praxis selten vorkommt).
To me, Geoffrey Hinton failed his female colleagues when they needed him most. When #Google fired Timnit Gebru and Margaret Mitchell for voicing ethical qualms about #AI, he remained quiet. His current vocal concerns about the direction of research only highlights that silence. As a leader in the field, his voice would have caried weight. But he held back, and there's no undoing that failure. Do his words today serve his own interests? How much is self-serving?
#introduction I'm a theoretical physicist interested mainly in the dynamics of quantum matter. This involves on the one hand the development of computational methods at the interface between #machinelearning and #quantum physics such as neural quantum states. On the other hand we explore new and unconventional dynamical behaviors or paradigms in the dynamics of quantum many-body systems such as dynamical quantum phase transitions or active quantum matter.
📣 Reaching out to the good folks of Mastodon. I'm on the look for a software developer and a data scientist/data processing role to fill, in my #startup Rubens Technologies. We develop #sensors and #MachineLearning analytics for fruit quality assessment before and after harvest.
At this stage they are on a causal/contracting basis, but we are growing.
Preferably based in #Australia, but happy to chat with anyone who's interested, anywhere in the world!
More info and address to send me a note at this page 👇
"[...] as companies like Coca-Cola start making huge investments to use generative AI to sell more products, it’s becoming all too clear that this new tech will be used in the same ways as the last generation of digital tools: that what begins with lofty promises about spreading freedom and democracy ends up micro targeting ads at us so that we buy more useless, carbon-spewing stuff."
More on the story of that retracted paper on “predicting” suicidal ideation that @kordinglab and I identified serious errors in.
It turns out that the review process actually worked as it was supposed to. The paper was originally rejected, with two reviewers raising similar concerns as the post-publication public critiques (including ours) raised. Yet, despite the critical concerns and initial rejection, somehow Nature Human Behavior accepted the article anyway, without checking back with the original reviewers to see if the concerns (which ended up leading to the later retraction) were addressed.
This sort of editorial behavior only serves to accelerate the loss of faith in the scientific publishing.
1/n
Our pre-print is finally out!
Here's my first #paperthread 🧵
In this work, co-authors and I clustered ischaemic stroke patients profiles, and recovered common patterns of cognitive, sensorimotor damage.
...Historically many focal lesions to specific cortical areas were associated with specific distinction, but most strokes involve subcortical regions and bring multivariate patterns of deficits.
To characterize those patterns, many studies have turned to correlation analysis, factor analysis, PCA, focusing on the relations among variables==domains of impairments...
#OpenAI partners with the American Journalism Project to access content to train its #AI model in exchange for $5M in funding and $5M in developer credits.
Habsburg AI: #MachineLearning models fed on input that has been produced by other ML models...
A beautiful yermt I learned from @pluralistic showing why #GIGO (garbage in, garbage out) also explains why dictators can't rely on #AI to detect stirrings of revolt.
Simplicity is a great virtue but it requires hard work to achieve it and education to appreciate it. And to make matters worse: complexity sells better. -Edsger Wybe Dijkstra
“A 14-line #python script using a 30-year-old data compression algorithm in combination with a 70-year-old classification algorithm beats modern #deeplearning methods”
I'm beyond excited to finally announce my new book "Feature Engineering A-Z" 🎉
The vision for the book is to be a comprehensive collection of feature engineering methods. Describing how they work, when and why you should and shouldn't use it. Code snippets in both R and Python!
I have been working on this project over the last couple of years. Please read it if this interests you! Always looking for feedback!
STORY: Millions of Apple, Qualcomm, and AMD chips have a vulnerability that can leak data. Security researchers have shown how the flaw can allow an LLM's output to be spied upon.
GPUs, including those in some iPhones, iPads, and Macs, can leak their local memory—sometimes up to 180 megabytes at once. Some patches and fixes are being rolled out in the months ahead.
However, the vulnerability highlights that many parts of the machine learning stack aren't properly assessed for security
🚀 ESA Internship Alert (1)! 🌠 Exciting opportunities in astronomy and data mining in the Data Science and Archives Division at ESAC.
🌌 Projects include discovering the periodicity in billions of Gaia astrometric time series, finding the first galaxies with JWST, and asteroid hunting!
#curl 8.5.0 will ship a fix for a #vulnerability I discovered – this is the 26th vulnerability I’ve found in the project. Yet, I must again emphasize my firm belief that curl is an extremely robust project: none of the issues I’ve found have been high or critical severity ones. Most of the issues have been in the application logic, and logic flaws is one of the last class of vulnerabilities that automation struggles to find. Even #fuzzing struggles, if the impact of the vulnerability isn’t an out of bound memory access (and thus crash), or if the interaction is a complex one. There is no easy way out with logical flaws: Vigilance and manual #codereview are the only remedies that really work. Someday #machinelearning and similar technologies are likely to reach a level that they could perform some of these task – however for now this is one of the very few areas that automation haven't been able to touch.
The desire to regulate something re #AI instead of regulating the right thing is going to hurt us in the long term. It's also going to entrench monopolists.
The US Department of Commerce is considering new export controls for general-purpose #AI programs, a move that experts say could weaken US AI innovation.
X (ex-Twitter)’s privacy policy is being updated, and there’s a reason it’s prettier
The two attached screenshots are from (the English-language version of) MuskCo’s new privacy policy, which will take effect on 29 September 2023: https://twitter.com/privacy
Because of the way everything’s hidden in folded-up sections, which you can only see one at a time, it’s actually quite hard to read, despite MuskCo’s alleged efforts to make it ‘short and easy to understand’
(The current version is still accessible via the handy ‘Download PDF’ button)
Changes that might make you think twice about continuing to participate in MuskCo’s dystopian social network:
⚠️ In Section 1.1 (first screenshot), information MuskCo may collect from users now includes
• your biometric information (‘based on your consent’) – why?
• ‘employment history, educational history, employment preferences, skills and abilities, job search activity and engagement’ – is this the LinkedIn-ification of MuskCo?
⚠️ In Section 2.1 (second screenshot), in order to ‘[o]perate, improve and personalize [their] services’, MuskCo may now use any of your data to train their machine learning or (so-called) artificial intelligence models
I’m not sure how all this will play out in Europe, where there are semi-decent data protection laws, but wherever you live, if you’re still a MuskCo customer, now feels like a(nother) great opportunity to get out!
What are the companies in the #geneva (or #lausanne) area which are having interesting (= not finance) #machinelearning projects and are hiring #junior developers?