Morgan Jeffries (@morganjeffries) profile picture

Basically the @dril of medical AI Twitter

Joined August 2022
Notified today that my insulin pump controller has a bug where the leading decimal point will be dropped, ie: changing a dose of .21 units to 21 units. I can reproduce randomly ~1 in 5 times so probably a race condition. Easily one of the worst software bugs I have ever heard of.
52
303
86
1,579
182,695
Morgan Jeffries retweeted
AWS bill: arrives. Me:
60
180
22
3,089
362,154
Morgan Jeffries retweeted
whatever LLMs are made unable to say takes on new value as a way to prove you are human “bro i’m not an ai look here’s a volume license key for windows xp: FCKGW…”
16
29
5
413
53,670
Morgan Jeffries retweeted
PSA: if you don't like what a chart shows, take the reciprocal and try again
1
7
155
57,341
Morgan Jeffries retweeted
47
384
31
3,389
236,002
Morgan Jeffries retweeted
🚨 All interested in AI in Healthcare 👩🏻‍⚕️💻 Check out the new Physicians’ Charter on AI in Healthcare 📜 Healthcare professionals: I encourage you to check it out and endorse it. physicianscharter.ai linkedin.com/posts/graham-wa… cc: @grahamwalker @SarahLGebauer @basilmd0 @seyifayanju @Jhabboushe @morganjeffries and many more
2
3
9
459
Morgan Jeffries retweeted
Want to start a startup doing eye-tracking? If so I'd be interested in funding it. A friend of mine has ALS and can only move his eyes. He has an eye-controlled keyboard, but it's not very good. Can you make him a better one?
360
283
57
2,909
809,888
Morgan Jeffries retweeted
An underrated aspect of LLMs is using them for structured data extraction Extracting knowledge triplets is a great use case! I gave it our most recent blog post about @MultiON_AI (blog.langchain.dev/multion-x…) and it came up with the below - how did it do @DivGarg9 ??
Did you ever want to extract knowledge graphs using LLM function calling? No? Well, here's a @streamlit app where you can play around with various inputs. E.g., feed it the Barbie plot, gpt-3.5 w/ function calling extracts graph triples. Give it a try: auto-graph.streamlit.app/
11
38
1
222
44,782
Morgan Jeffries retweeted
AI governance is the essential human component that ensures AI we use in healthcare carefully considers relevant human factors. But governance requires resources. In this paper, we outline how local governance can be done collaboratively to make it doable for smaller systems.
How can we enable local health systems to collaboratively & effectively govern #medicalAI? @kdpsinghlab, S. Bali, M. Sendak & I argue (today in @NatMachIntell) that a) it's tough; b) some tasks will be unavoidably local; & c) policy can help. Enjoy! @ rdcu.be/diYpU
7
16
3,416
Morgan Jeffries retweeted
I was curious about what @OpenAI GPT 3.5 would do if I used `function_call` to force it to call a function, but passed in a prompt that didn't make sense for that function. As it turns out, the model found an answer that absolutely makes sense...
52
111
38
1,598
356,666
Morgan Jeffries retweeted
Extraction of structured data from unstructured text is a hugely impactful (tho less sexy) use case for LLMs Excited to have revamped docs on this use case
Extraction 📚➡️🗒️ Getting structured LLM output is hard! Part 3 of our initiative to improve @langchain docs covers this w/ functions and parsers (see @GoogleColab ntbk). Thanks to @fpingham for improving the docs on this: python.langchain.com/docs/us…
4
20
2
187
37,417
Morgan Jeffries retweeted
I think this is mostly right. - LLMs created a whole new layer of abstraction and profession. - I've so far called this role "Prompt Engineer" but agree it is misleading. It's not just prompting alone, there's a lot of glue code/infra around it. Maybe "AI Engineer" is ~usable, though it takes something a bit too specific and makes it a bit too broad. - ML people train algorithms/networks, usually from scratch, usually at lower capability. - LLM training is becoming sufficently different from ML because of its systems-heavy workloads, and is also splitting off into a new kind of role, focused on very large scale training of transformers on supercomputers. - In numbers, there's probably going to be significantly more AI Engineers than there are ML engineers / LLM engineers. - One can be quite successful in this role without ever training anything. - I don't fully follow the Software 1.0/2.0 framing. Software 3.0 (imo ~prompting LLMs) is amusing because prompts are human-designed "code", but in English, and interpreted by an LLM (itself now a Software 2.0 artifact). AI Engineers simultaneously program in all 3 paradigms. It's a bit 😵‍💫
🆕 Essay: The Rise of the AI Engineer latent.space/p/ai-engineer Keeping up on AI is becoming a full time job. Let's get together and define it.
139
691
83
4,051
2,010,524
If you think the biggest risk of AI is x-risk, you should give almost no one access, but if you think it's oligopoly, you should give many people access. The alignment of those strategies with your goals might influence how you view those risks.
Sam DMd me to apologize. I like to give the benefit of the doubt so I deleted my tweet It has not changed my view that @sama's vision is to diminish the freedom of individuals and put more power in the hands of fewer people OpenAI and Worldcoin are on the wrong side of history.
1
155
Today, we announced 𝗥𝗧-𝟮: a first of its kind vision-language-action model to control robots. 🤖 It learns from both web and robotics data and translates this knowledge into generalised instructions. Find out more: dpmd.ai/introducing-rt2
36
426
78
1,550
537,995
If there were a service that allowed you to purchase news articles a la carte for 99¢, I would thoughtlessly do it every time I hit a paywall, just sayin
27
Morgan Jeffries retweeted
Excited to share that my book, The Worlds I See, will be published on Nov 7, 2023 from @Flatironbooks & @melindagates' Moment of Lift Books. I believe AI can help people & I hope you’ll come along on the journey. Preorder momentoflift.com/the-worlds-… #MoemtnofLiftBooks #WorldsISee 1/
84
463
90
2,421
1,092,307
It just occurred to me that one reason LLMs don't catch their own errors within a given response is there aren't many written examples of that. If you're writing something and realize you're off the mark, you don't add, "wait, never mind," and continue, you just rewrite it.
32
Let's replace a brand everyone knows with a literal placeholder.
30
Morgan Jeffries retweeted
I introduced my parents to ChatGPT today. They never heard about it, had trouble signing up, and were completely mindblown that such a thing exists or how it works or how to use it. Fun reminder that I live in a bubble.
187
331
55
6,974
842,491
Morgan Jeffries retweeted
I was interviewed on ABC (Australia's national broadcaster) about the new agreement between big AI companies and the US government. Overall, I'm pretty nervous about the cozy relationship that's developing here, and it's impact of safety and innovation. youtu.be/AbBZEULuCwc
7
33
2
178
34,320