Companies seemed to realize that "I want to talk to a human" is used all too often and they put up more barriers before the bot actually connects to customer support

This is where I lose my patience with the whole thing, after the 3rd time the agent blocks me from doing it, as a paying customer

A low bar is whether public procurement officials understand the significance of a question like, “What’s the precision-recall AUC of the computer vision classification?”, otherwise taxpayers should not be footing the bill

I wonder how Mistral OCR compares for this kind of task. I recall it having better accuracy on benchmarks relative to Gemini at release

mistral.ai/solutions/do...

For example, Cathie Wood is one of those people who might have a good eye for business but every time I hear her explaining a basic AI concept on a podcast, I'm compelled to turn it off

I use it mainly for formatting notes (e.g. turning lists to tables or turning paragraphs into lists), further developing my half-baked thoughts, and scaffolding folders and (re)organizing notes with updated tags via prompts

I love how extensible Obsidian is! One thing that I've recently started using is the Claude MCP extension for Obsidian. Basically, you can give Claude permissions to search and edit your vault. I use it to help rewrite and organize notes: github.com/StevenStavra...

Exactly- the biggest problem I've seen with this in practice, and one that this article misses, is that it assumes govt decision-makers have a complete and thorough grasp of algorithmic tools and their limits. I've yet to meet a policy person with an effective sociotechnical understanding of AI.

Claude catches on quick

Regulations aim to associate costs with unethical or negligent decisions. They are important in protecting the health and safety of the public. In light of this example, perhaps the state needs to be doing more to regulate AI development to mitigate the potential of public harm?

In addition to the harm the system presents from misalignment and bias amplification, framing "uncensored" LLMs as those without appropriate safeguards and labeling them "good" is a failure of ethical development practices and should come with stronger repercussions.

I've read and watched many resources to keep up with AI, what initially helped me bridge machine learning concepts with LMs/genAI was The Little Book of Deep Learning by @francois.fleuret.org. Highly recommend it for anyone in the same position: fleuret.org/francois/lbd...

Chris Hayes' new book, The Siren's Call, is all about this concept. There's a good discussion of it here: podcasts.apple.com/us/podcast/a...