8 April 2023

Why do I blog?  Danny Guo lays out his reasons.  For me, it is to document the salient items of what's interesting to me, and perhaps it might be of interesting to a passing visitor.  It is a journal of sorts.

––– 凄い –––

Survival of the fittest.  Stability AI, which took the world by storm with the Stable Diffusion AI image generator, now has been beaten by competitors who were lucky, or just ran faster with the ball, especially Midjourney.  Now, they are hard pressed to find investors and may have to wind down.  Harsh is the world of competitive geekdom.  But Midjourney does seem to be better.

––– 良くない –––


––– 良くない –––

Blame Obamacare.  This is a great article on how government policies greatly changed the face of medicine for the worse.  What resulted from Obamacare was the eradication of the small doctor practice, and the rise of the employed physician.  Insurance companies ruled the land, and so entities were created to help them make profit, like pharmacy benefit managers.  A whole industry came into existence to help healthcare systems deal with insurance billing and other administrative overhead.  Hospitals also became dominant as they received generous largess from Washington D.C.  The glass and steel buildings on the Hill and the waterfront were obtained during Obama's tenure.  Big Medicine didn't help doctors, however.  During that time, doctors sold their practices and became employed physicians, and what care you received was no longer up to the doctor, but what the insurance company would cover.  Most physicians may struggle to pay off their debts now.  Medicine will become more like a government-run operation (which it kind of is), than a place for holistic healthcare. 

––– 良くない –––

COVID is over, right?  So why does there need to be a special ICD-10 code for COVID-19 jab status?  We have never had a special code for who didn't take a medication, or who didn't undergo a certain surgical procedure.  Although the Z28 codes for not being vaccinated was created in 2016, the code has been fine-tuned specifically for the COVID-19 vax, to document not only that a person did not get the jab, but the basis for this status (e.g. group pressure, refusal etc.)  The government says that it needs this to track vaccine efficacy and disease mortality, but they never were interested in this before with other diseases.  So why now?
Update (9 April 2023):  Here is the reason: it "will help health insurance providers identify emollees [sic] who may benefit from outreach and further education about vaccination.”  
––– 良くない –––

Red flag law update.  In late December 2022, a New York Supreme Court judge declared their "red flag" gun law unconstitutional.  Then recently, a person was able to use this decision to vacate and dismiss a red flag judgment, affirming the prior decision (R Monaco vs State of New York).  I never understood why it became illegal to defend yourself in the most effective way, in a state with high crime and violence.  

––– 凄い –––

Are you nuts, folks?  Seattlites are all for confiscating guns from law-abiding citizens, but reject an opportunity to address the real issue?

––– 良くない –––

A Harvard physician/computer science reported that he was able to get GPT-4 to diagnose a rare medical condition (congenital adrenal hyperplasia).  Of course, he was the prompt engineer, so his success might not be duplicated by someone else.  The article states:

GPT-4 isn't always reliable, and the book is filled with examples of its blunders. They range from simple clerical errors, like misstating a BMI that the bot had correctly calculated moments earlier, to math mistakes like inaccurately "solving" a Sudoku puzzle, or forgetting to square a term in an equation. The mistakes are often subtle, and the system has a tendency to assert it is right, even when challenged. It's not a stretch to imagine how a misplaced number or miscalculated weight could lead to serious errors in prescribing, or diagnosis. 

Like previous GPTs, GPT-4 can also "hallucinate" — the technical euphemism for when AI makes up answers, or disobeys requests.

So it's not yet time to ditch your physician for an A.I. model.  And see the first item from my 6 April 2023 posting.

––– 凄い –––

The first time [Priya] received her paycheck, it was the most money she’d ever seen in one place. Uttar Pradesh is infamous for having higher school dropout rate among girls compared to other parts of the country. Its not exactly the safest place for girls either. Her father had invested in her education and kept her in school. And it had paid off. Sending your first salary back home is one of the proudest moments in a young Indian’s life, and that was the case for Priya also.

I think you know where this is headed. The day I got off the GPT4 wait-list I asked it to do what Priya does on a daily basis. It got the answer wrong in the first try, but some chain-of-thought prompting and boom. GPT4 gave the correct answer in 1/10th the time it would take Priya and cost a lot less.

I don’t see a long-term career in software anymore. Any dreams I had of earning decent money as a software engineer are slowly fading. Lex Fridman in his podcast said “if you’re anxious about GPT4 its probably because you’re a shitty programmer”. I mean, I’m not the smartest in the room but I have generated value with the software I’ve made. And was convinced that I’ll make decent money as long as put in the work. I’m just not that sure anymore.

What is the economic impact of LLMs? Idk (openAI has published some lengthy paper about it). What I do know is that some rich bloke in the US will get a few million dollars richer and Priya will lose her job.

––– 良くない –––

So why are we wearing masks?  It's 2023, and only now we're learning that coronavirus present on hands and other surfaces can spread COVID-19.  Everyone touches their masks, and then touches everything else in the grocery store.

––– 良くない –––