No system is stronger than its weakest link, and there are so many points of failure for "AI".
I have reimagined a diagram from Suresh & Guttag, noting where the weak links in machine learning are, both in data generation as well as model building and implementation.
axbom.com/bias-in-mach...
Tagged: Bluesky
↩ Replying to a post
Thank you Sarah! Just dipping my toes in again and testing the waters. It's really nice to see so many friendly faces from the 'olden times' 😅
↩ Replying to @shaungomm.bsky.social
Thank you Shaun! We're already reaching that age apparently. Good to hear you got the help you needed. We'll have to meet up and have a heart-to-heart! 🥁🫶
The blog post I wrote in Swedish the night before my open heart surgery is also available in English.
In an addendum, I made note of how overly excited I was to 'change things' before surgery, and how exhausted I am now.
A large part is an hommage to Lillian Gilbreth. 😊
axbom.com/heart-surger...
When I first published my poster "If a hammer was like AI" (more than a year ago) many people considered the carbon cost point to be too much of an exaggeration. I don't get as much pushback on that point anymore.
axbom.com/hammer-ai/
An explanatory model for the relationships between different actors in AI development and use. Who finances the systems, who contributes to the systems and who benefits or suffers from them? The article contains explanations of each grouping in the chart. #AIEthics
axbom.com/aipower/
One of my predictions is that more and more employees will start writing down texts that argue for their own vision to be realised, but they will write it as if it was an AI output.
Because they know management will listen to an AI but not to them.
"From the generic hipster cafe to the ‘Instagram wall’, the internet has pushed us towards a kind of global ubiquity – and this phenomenon is only going to intensify"
www.theguardian.com/news/2024/ja...
The tyranny of the algorithm: why every coffee shop looks the same
A bit too many networks and platforms again. This is the one I have ignored the most.
What would you consider are the benefits of Bluesky?
Certainly not the level of engagement.
Zero responses in a month and a half… 😂 I believe that's the worst I've seen since 2004.
« [Julian Joseph] used a tool called LazyApply to blast out 5,000 job applications "in a single click" — and managed to land around 20 job interviews. While that may sound a lot, that's a meager success rate of roughly half a percent. »
Gotta love the future.
futurism.com/the-byte/une...
Unemployed Man Uses AI to Apply for 5,000 Jobs, Gets 20 Interviews
Tech criticism does not stifle innovation. It influences public opinion and ENCOURAGES INNOVATION that protects and contributes to well-being.
The trope that ethics and sustainability issues are tech-hostile is getting old.
All industries need criticism to improve and innovate.
#DigitalEthics
A common argument I come across when talking about ethics in AI is this one: it’s just a tool, so "it's really no different from a hammer".
So I made this poster. You can find more context for the poster’s topics in my blog post: axbom.com/hammer-ai/
#AiEthics #DigitalEthics
Diagram: Bias in Machine Learning
Understand the stages of machine learning where bias can, and often will, contribute to harm.
#DigitalEthics #AIEthics
axbom.com/bias-in-mach...
The Elements of Digital Ethics
A chart to help guide moral considerations in the tech space.
axbom.com/elements/
#DigitalEthics