Konrad Kording

The Electric Motor and the Drill - we use AI in the wrong way Power tools are better than general purpose tools for most applications, my science planning app planyourscience.com is a result of this philosophy

In which @kordinglab.bsky.social argues LLMs are more like an electric motor than a drill, and starts to build a drill for scientific research.

open.substack.com/pub/kording/...

The Electric Motor and the Drill - we use AI in the wrong way Power tools are better than general purpose tools for most applications, my science planning app planyourscience.com is a result of this philosophy

New post: The Electric Motor and the Drill
LLM chat bots are like electric motors: can do anything, bad at everything. The problem is UI/UX. AI should be more like power tools. I built a science planner around this philosophy. 10,000 scientists use it now. open.substack.com/pub/kording/...

I'm biased, but I think Neuromatch has done a lot of good in helping make computational science education accessible. Please consider donating to help us continue!

How I contributed to rejecting one of my favorite papers of all time I believe we should talk about the mistakes we make.

How I contributed to rejecting one of my favorite papers of all times, Yes, I teach it to students daily, and refer to it in lots of papers. Sorry. open.substack.com/pub/kording/...

PlanYourScience.com is live! 🔥

Very cool. Which proportion of cortical neurons do you think would show similar effect sizes?

I am with @argalloni.bsky.social . If the literature is convincing giving him a link will cost you 30 seconds. If it is not, asking you for a reference avoids the “princess in a different castle” phenomenon where the weakness he spots is ignored in debate.

Wonderful! More science is needed!

I would be shocked if our world models materially change how we feel.

I never understood why sleeping was hard. I put my head down. And am out. Pretty much every time. Obviously provided I am tired enough.

May suggest that my own inner space colored my questions! Very interesting.

I don't know. Mine became explicit very quickly once I started asking myself during thinking where I was and which capabilities I had in that simulation.

very cool!

World-models in your head Talking with a lot of people, they have rather shocking different kinds of world-models. I believe that people have somewhat specialized simulators. Let me list some and then give you the chance to ad...

Let's compare our world models. I find that different people seem to have rather distinct internal world models. E.g. I personally have neither visual imagination nor an inner voice, found it weird others do. Here is a quick google forms to check idea:
docs.google.com/forms/d/e/1F...

It clearly exists, there are neurons known to sense hunger.

yes, a super cool paper!

It is called biofeedback. Its a thing. You take activities of something. You display it to the animal or person and reward them for making the activity go up or down.

Fine. Fine. I meant cortex. And was rather referring to a somewhat broad pattern instead of meaning actually anywhere. Sorry.

Ah good point!

There are neurons that directly sense “hunger”…

Interesting. I would have thought that is not about having hunger but the response to it. (I know preciously little about this).

I am recruiting graduate students for the experimental side of my lab @mcgill.ca for admission in Fall 2026!
Get in touch if you're interested in how brain circuits implement distributed computation, including dopamine-based distributed RL and probabilistic representations.

Interesting mystery. It is known that animals can learn to control neurons pretty much anywhere in the brain. But they can not learn to ignore hunger which probably means they can't turn of hunger sensing neurons. How is that avoided in the brain? They even have DA inputs.

thanks a ton!

exactly. Its just I don't seem to be able to find a clean paper on that.

VTA DA neurons. I read they have heterogenous projections to cortex. And they have different tuning curves in terms of Reward, punishment, hunger, etc. Computationally it would make a lot of sense that the relevant reward signals make it to the right cortical areas. Known? Computationally important.

Nice. Did not know this! very cool!

That is a fun framing in which rates can’t be formulated. I can do that for the other side. In reality, there is time, say at MHz resolution. In every bin there is either a spike or no spike. So clearly only rates matter ;)

I did not think I was caving. There are obvious places where spike counts matter - e.g. neuromuscular junctions. There are obvious places where timing matters (auditory system). I said "rates only matter" was clearly wrong. Not that oscillations are hugely important - I think we don't know that.

Professors: Running a class with research projects? You and your TAs will never have enough time to carefully guide every student through gap, hypotheses, experiment, data analysis, etc. I think every larger course running research projects should use this.