Yeah, medical computer vision is a (fascinating) field with a lot of ongoing research. SOTA models are highly specialized, and are only getting good enough to be used by actual doctors and patients. Using a general purpose LLM to do this is similar to giving a credit card to Openclaw and telling it to make you rich through the stock market & cryptos.
Doesn't track with mine. I've been stuck with Sonnet 4.6 with one of the clients I work for. It writes code fine, but it's not nearly as good as the more recent models for everything else. It's fairly common for it to suddenly go off the rails for no good reason, so I can't really trust it with agentic loops.
It's also not very good at diagnosing non-trivial issues. It's not uncommon for it to suggest whole lists of irrelevant / nonsensical reasons for something not working. Then I copy/paste the code and some context into chatgpt and it hones in onto the correct issue right away, even with inferior tooling.
Having to lug 2 phones around has always seemed like more trouble than it's worth to me. I also don't like having multiple devices to do stuff that a single one could do, for environmental reasons, but that's not a very wide-spread opinion.
So I do have work stuff on my personal phone, but with no notifications whatsoever. Only works because I'm in a position where it's acceptable to require all communications to go through emails or messaging apps though.
Can't speak for 26, but a year ago I worked on a project that migrated from v5 to 11 because of improved image segmentation capabilities. My understanding is that the newer versions don't necessarily have better precision/recall, but they tend to be faster for equivalent results, and have increased capabilities.
Had a very similar experience. Opus went "look, t-sne shows your features are neatly clustered" (it didn't) and left it at that. Fable didn't fully explore the problem/data, but it did go much further, implementing models to check for correlations and adjust feature clusters. Opus was able to finish the job after Fable was cut, but required much prodding (doing exactly what you described: pointing it towards things that look off and asking it, are you sure that's all there is to this?).
Useful context for this is that token usage keeps rising at an exponential pace. I mean, we don't have numbers for the big labs, but Openrouter's numbers are quite telling (can't post link because corporate decided to block all "non-validated AI tools"), and I think they're probably representative of the global trend. +500% year to date, +50% over the month of May alone. It's unsurprising that providers are struggling to find and pay for the compute.
> In Bitcoin you don't generate cash, you earn block rewards for acting as a consensus broker which otherwise would require a central banking settlement layer. This activity, tied directly to the transaction layer, acts to maintain the equilibrium between increases in goods and services and expansion of the money supply.
Block rewards have no connection to transaction volume or economic activity, the protocol is designed such that bitcoin supply increases at a predictable (and diminishing) rate. Bitcoin is deflationary by design, which is one of the major issues that stopped it from becoming anything other than a speculative store of value.
> at it's core (pun intented) Dark matter is something to make equations fit without any other thought behind it or whether there might be several things behind it or god forbid that we juddge the equations themselves
Another way to interpret dark matter is that we can observe something using several different ways, but all those ways use gravity. When trying to observe this something using electromagnetism, we see nothing. It doesn't seem so crazy then to hypothesize that this something only interacts with gravity, and not electromagnetism.
Not GP, but BMAD has several interview techniques in its brainstorming skill. You can invoke it with /bmad-brainstorming, briefly explain the topic you want to explore, then when it asks you to if you want to select a technique, pick something like "question storming". I've had positive experience with this (with Opus 4.7).
As a data architect I dislike the term NoSQL and often recommend that my coworkers not use it in technical discussions, as it is too vague. Document, key-value and graph DBs are usually considered NoSQL, but they have fairly different use cases (and I'd argue that search DBs like Elastic / OpenSearch are in their own category as well).
To me write scaling is the main current advantage of KV and document DBs. They can generally do schema evolution fairly easily, but nowadays so can many SQL DBs, with semi-structured column types. Also, you need to keep in mind that KV and document DBs are (mostly) non-relational. The more relational your data, the less likely you are to actually benefit from using those DBs over a relational, SQL DB.
> Certainly can’t compete with using VS Code or TeXstudio locally, collaborating through GitHub, and getting AI assistance from Claude Code or Codex.
I have a phd in economics. Most researchers in that field have never even heard of any of those tools. Maybe LaTeX, but few actually use it. I was one of very few people in my department using Zotero to manage my bibliography, most did that manually.
Money is never backed by nothing, or it's worthless. It may not be backed by anything physical, but it's always backed by some form of trust. National currencies are backed by trust in the corresponding government and institutions.
Reserves matter even if reserve ratios are zero. If Bank A lends too much money, then when its customers spend that money, a lot of it will end up deposited at other banks. These banks will then ask Bank A for reserves (as in, central bank money) to clear the inter-bank transfers, which Bank A will need to borrow from the central bank, at a cost.
This probably won't make you feel any better, but banks don't really loan out money that's not theirs. When they lend money, they literally create it out of thin air. Creating that money has a cost, which is what ultimately limits how much they can lend, and having more deposits can lower that cost somewhat, but there's no direct connection between the money you deposit in your account and the money that the bank lends to someone else.
If I were to do this (and I might give it a try, this is quite an interesting case), I would try to run a detection model on the image, to find bounding boxes for the planets and their associated text. Even a small model running on CPU should be able to do this relatively quickly.
On the professional side, they also often let you interact with their experts and architects directly, as part of your support contract. With most other companies, you either have to go through front-office support exclusively, or pay extra for Professional Services.
> I’m downplaying because I have honestly been burned by these tools when I’ve put trust in their ability to understand anything, provide a novel suggestion or even solve some basic bugs without causing other issues.?
I've had that experience plenty of times with actual people...
LLMs don't "think" like people do, that much is pretty obvious. But I'm not at all sure whether what they do can be called "thinking" or not.
> In the examples given, it’s much faster, but is that mostly due to the missing indexes? I’d have thought that an optimal approach in the colour example would be to look at the product.color_id index, get the counts directly from there and you’re pretty much done.
So I tried to test this (my intuition being that indexes wouldn't change much, at best you could just do an index scan instead of a seq scan), and I couldn't understand the plans I was getting, until I realized that the query in the blog post has a small error:
> AND c1.category_id = c1.category_id
should really be
> AND p.category_id = c1.category_id
otherwise we're doing a cross-product on the category. Probably doesn't really change much, but still a bit of an oopsie. Anyway, even with the right join condition an index only reduces execution time by about 20% in my tests, through an index scan.
Interestingly, "aggregate first, join later" has been the standard way of joining fact tables in BI tools for a long time. Since fact tables are typically big and also share common dimensions, multi-fact joins for drill-across are best done by first aggregating on those common dimensions, then joining on them.
Makes you wonder how many cases there are out there of optimizations that feel almost second nature in one domain, but have never been applied to other domains because no one thought of it.
> - It says it's done when its code does not even work, sometimes when it does not even compile.
> - When asked to fix a bug, it confidently declares victory without actually having fixed the bug.
You need to give it ways to validate its work. A junior dev will also give you code that doesn't compile or should have fixed a bug but doesn't if they don't actually compile the code and test that the bug is truly fixed.
Properly measuring "GPU load" is something I've been wondering about, as an architect who's had to deploy ML/DL models but is still relatively new at it. With CPU workloads you can generally tell from %CPU, %Mem and IOs how much load your system is under. But with GPU I'm not sure how you can tell, other than by just measuring your model execution times. I find it makes it hard to get an idea whether upgrading to a stronger GPU would help and by how much. Are there established ways of doing this?
Open source models like Flux Kontext or Qwen image edit wouldn't refuse, but you need to either have a sufficiently strong GPU or get one in the cloud (not difficult nor expensive with services like runpod), then set up your own processing pipeline (again, not too difficult if you use ComfyUI). Results won't be SOTA, but they shouldn't be too far off.
I pay for chatgpt because, in my experience, o3 and o4 are currently the best at combining reasoning with information retrieval from web searches. They're the best models I've tried at emulating the way I search for information (evaluating source quality, combining and contrasting information from several sources, refining searches, etc.), and using the results as part of a reasoning process. It's not necessarily significant for coding, but it is for designing.
> Besides protein folding, the canonical example of a scientific breakthrough from AI, a few examples of scientific progress from AI include:1
> Weather forecasting, where AI forecasts have had up to 20% higher accuracy (though still lower resolution) compared to traditional physics-based forecasts.
> Drug discovery, where preliminary data suggests that AI-discovered drugs have been more successful in Phase I (but not Phase II) clinical trials. If the trend holds, this would imply a nearly twofold increase in end-to-end drug approval rates.
> It’s insane to me that maybe every bank I use requires SMS 2FA, but random services I use support apps.
It never ceases to surprise me how much American banks always seem to lag behind with regards to payment tech. My (european) bank started sending hardware TOTP tokens to whoever requested one like a decade ago. They've since switched to phone app MFA.