Jonathan Sweeney retweeted
Hivemind just completed its first swarm exercise with Taiwan's NCSIST!
In under three months, NCSIST engineers, supported by Shield AI, built AI pilots and put them on three Mighty Hornet III drones. Hivemind launched, coordinated a wide-area search across separate zones, kept the aircraft in communication with each other, worked around a no-fly zone, and landed all three once the mission was done. The exercise completed NCSIST's original contract and expanded it to put Hivemind on more of NCSIST's unmanned systems.
"Shield AI's hands-on training and technical expertise gave Taiwanese engineers the confidence to integrate and use Hivemind on our own." - Li Shih-chiang, NCSIST President
Full release: bit.ly/4q1OcOz
Autonomy for the world. The greatest victory requires no war.
Jonathan Sweeney retweeted
Chief Justice Roberts: AI won't replace lawyers, but legal research will become "unimaginable" without it He's right on both counts but this understates how radically the legal profession will be transformed. Lawyers have been insulated from disruption, but this isn't a healthy way for an industry or society to progress bloomberg.com/news/articles/…
Jonathan Sweeney retweeted
The whole ethic of America, by contrast, is “live fast die young.”
Insane to let a country with these stats build civilian airplanes. We’re here to innovate not to get you home safely!
Jonathan Sweeney retweeted
An early look at how the OpenAI Retrieval API works 👇
A few thoughts on how OpenAI is implementing RAG in the new Assistants Retrieval tool before I was locked out.
1. They're splitting on newlines. You can tell because they forget to insert the newlines back between the splits when giving you the reference (red squiggles).
2. The reference is contiguous. I haven't tried more complicated questions so can't rule out that they would do some kind of subquestion decomposition, but the queries I tried all just gave a single reference.
3. The number of chunks returned is variable. This indicates some kind of "small to big" strategy where they retrieve more context than they match. Perhaps something similar to our AutoMergingRetriever docs.llamaindex.ai/en/stable… w/ some kind of similarity cutoff but worth a bit more investigation.
4. They have some trouble handling UTF-8.
In general, very bleeding edge still, but look forward to seeing how this evolves.
Jonathan Sweeney retweeted
Laws to ensure AI applications are safe, fair, and transparent are needed. But the White House's use of the Defense Production Act—typically reserved for war or national emergencies—distorts AI through the lens of security, for example with phrases like "companies developing any foundation model that poses a serious risk to national security."
Yes, AI -- like many technologies such as electricity and encryption -- is dual use in the sense it can be used for civilian or military purposes. But conflating AI safety for civilian use cases and military applications is a mistake.
It’s also a mistake to set reporting requirements based on a computation threshold for model training. This will stifle open source and innovation: (i) Today’s supercomputer is tomorrow’s pocket watch. So as AI progresses, more players -- including small companies without the compliance capabilities of big tech -- will run into this threshold. (ii) Over time, governments’ reporting requirements tend to become more burdensome. (Ask yourself: Has the tax code become more, or less, complicated over time?)
The right place to regulate AI is at the application layer. Requiring AI applications such as underwriting software, healthcare applications, self-driving, chat applications, etc. to meet stringent requirements, even pass audits, can ensure safety. But adding burdens to foundation model development unnecessarily slows down AI’s progress.
While the White House order isn’t currently stifling startups and open source, it seems to be a step in that direction. The devil will be in the details as its implementation gets fleshed out -- no doubt with assistance from lobbyists -- and I see a lot of risk of missteps. I welcome good regulation to promote responsible AI, and hope the White House can get there.
Jonathan Sweeney retweeted
Just watched an old @acquiredotcom interview with @justinkan and can't help but wonder if Atrium was just ahead of its time: would it have taken off in the GPT4 era?
Jonathan Sweeney retweeted
Our very own @m_held joins host and innovation evangelist @TomRaftery on this week's Digital Supply Chain podcast. Tune in to deep-dive with them into Fishtail's data-backed tech and hear some customer success stories never shared, ever before! https://teapawt.net/t.co/PJnaKUeCMg
Jonathan Sweeney retweeted
Introducing our Fishtail Head of October 🎉. Once a month, we will give a sneak peak into the awesome team behind (or in front?) of Fishtail. Introducing @sweenejo !
Read below about Jon's favorite project he is working on here at Fishtail.
Jonathan Sweeney retweeted
Boris facing investigation. Trump facing impeachment. Bibi facing indictment. Not a good week for right wing wannabe strongmen.
JUST IN: British Prime Minister Boris Johnson has been referred to a police watchdog over his alleged relationship with an American businesswoman when he was Mayor of London. cnn.it/2nlhUmb















