| > For me, success means "robust structural intuition". Perhaps frame it as understanding that's robust to adversarial noise? To fuzzing testing? There was a flurry of papers in the early 2000s that aimed to generalize biological robustness, borrowing from ideas and math from engineering. You might find these interesting: https://www-users.york.ac.uk/~lsdc1/SysBiol/kitano.robustnes... https://link.springer.com/article/10.1038/msb4100179 https://www.cs.unibo.it/~babaoglu/courses/cas02-03/papers/Ro... > An Ideal Gas Law chapter problem, with numbers for solid Argon - Ha! > First-tier institution students, coming to intro genetics from intro bio, lacking even a firm grasp of central dogma? Yeah, unfortunately this is a real problem: Foundational biology courses (intro, genetics, cell bio) overwhelm students with a firehose of facts that must be learned or you flunk out. Later, in upper-level undergrad and grad school, those facts start connecting, and biology becomes lots more interesting and actually easier to study. > Are labels and vocabulary treated as foundational, or as relatively unimportant? Vocabulary is a big deal in biology. Many terms carry associated meaning, for example polymerase chain reaction helps describe the mechanism, and TAQ polymerase reminds you that heat is important. Bone morphogenic protein says a lot. That said, plenty of biology terms are pretty useless. Ribosome doesn't provide much intuition other than RNA is involved, and Golgi apparatus is even worse. Many gene names are arbitrary, reflecting a lack of knowledge at the time of discovery. Some are just dorky like sonic hedgehog. Good undergrad biology books have big, carefully written glossaries in the back, these are absolutely invaluable. > How important is seeing why things are the way they are? It's important to internalize: 1) Biology is just physics and chemistry. 2) Millions of years of evolution and randomness produced all these arbitrary biological systems with their endless complexity. That's why living organisms are nothing like rational engineered systems, despite all the shared physics. > If you haven't seen part of the map, how important is being able to sketch it in plausibly? If fragmented into puzzle pieces, how important is being able to fit them together? For me, studying any big subject with lots of details, context really helps. It's easy for me to get lost in the details and lose motivation unless the ideas plug into some bigger picture. That's true even if I only want tourist-level knowledge. |
Regards such cross-cutting insights, I'm tempted to see what an LLM, given a order-100 author tome, where the expertise is applied mostly silo-wise, might make of something vaguely like "Edit this to emphasize modularity (see doc). And weave that story across chapters. And...". But I suspect a good breath-wise pass would require similar order massed expertise.
Hmm... just now doing a quick and sloppy spot check, suggests current AIs might be coaxed to draft a "simplified tree of life" using the regulatory miRNA family ratchet (and Hox clusters, TF families) as a lineage regulatory/complexity budget. That's potentially much easier to do than it's been in years past. So maybe the blocker of "massive expertise resources required, but there's little incentive", might be destabilized by AI?
> Later, [...] connecting, and biology becomes lots more interesting and actually easier to study
So my question is, can this be done much much earlier? A kindergarten sci ed person suggested their kids have a human right to understand their world now, not a lifetime (for them) later. Seems an intriguingly audacious goal. Riffing on above, might one create a K-accessible/empowering fun little tree of life?
Our collective focus seems elsewhere. From MCAT to early primary, I've heard "Yes, <that> would be a nice way to explain <concept>, providing better understanding. But my students are soon taking <next high-stakes exam>, and that's not on the exam. Our time together is limited, so I'd be doing them a disservice if I <didn't teach to the test>."
Perhaps the texts are fine and something extra is needed. A formative AI tutor? Or perhaps texts and tests could use a massive refresh. Or perhaps something non-textual is needed. Tens of hours living in novel AR cell sims? Or some combo. But I find the status quo rather piteous.
Viewing science education as disaster triage chain of care, there's a distinction between stabilization and patient packaging, just surviving handoff vs not having to redo. If intro genetics is burning time on something which could have been taught in primary but wasn't, and was taught unsuccessfully in middle school, again in HS, and again in intro bio... maybe change is needed and available? Order-of-magnitude size is a part of most every physical sciences intro class, and there was a NSF "nanotech" push to primary, so how many times and years might a first-tier med student have been "taught" it, 10-ish?, before having little clue how big a red blood cell is?
One teaches with the students, outcomes, and material at hand. So material which pervasively leverages scaling doesn't exist; and outcomes don't require/permit it; and students have no sense of scale; so such material would be undeployable; and sci ed research doesn't deal with the undepoyable; so we've no idea what it might look like or gain; so there's little research funding; and thus little progress over time. I think of us as being wedged, or in a local optimum with search temperature dialed unhelpfully low.
Sorry for my latency. Thank you again for the comments.