23 February 2026

Artificial intelligence agent takes your class for you

Making the rounds on socials today is a new generative A.I. agent that logs into your class website and takes the course for you. I have no inclination to promote this, so I am removing the name of the service, since it’s easily found if someone wanted to. 

This agent is supposed to allow a student to clock out from taking a course entirely. The description is worth reading.

How can [Agent] do all this?
[Agent] has his own virtual computer with a browser, just like you do. He can navigate websites, watch videos, read documents, type in text fields, click buttons, and submit forms. Anything you can do on a computer, [Agent] can do.

How does [Agent]  access my Canvas?
You link your Canvas account once during setup. [Agent] uses your credentials to log in, view assignments, and submit work on your behalf.

Will my professor know?
[Agent] submits assignments from your account just like you would. The work is original and generated per-assignment — not copied from a database.

As a professional educator, and one who has tried to give students a good online learning experience, you better believe that I have feelings about this. Quick notes.

Some students would never use this and recoil at the thought. We like those students.

Some students would try to use it for everything. Way back in 2020, Ian Bogost noted that for a lot of people, university is about an “experience” rather than an education. Maybe they are there to play sports, maybe they are there to party, maybe they are there hoping to find someone to marry. Passing classes while avoiding most or all of the work has always been possible for students with enough money to hire “essay writing services.” 

Some students would use it for some classes. And this is a group that I sympathize with. We educators should be thoughtful about why students might want to using this, even if they know they are cheating themselves out of learning. 

Some may have very specific interests and don’t see the value of taking courses that they see as unrelated to those interests. Some students would use if if they thought the workload was unreasonable or that the work boring. Some students would use it to not fail, particularly if they were going through a rough spot personally. We’ve loaded up a lot of consequences for students if they fail classes, not the least of which is financial aid.

I want taking a class to be so rewarding that the thought of giving that to someone else is as attractive to a student as a machine that offers to eat your food, drink your beers, watch your favourite movies, and go on dates for you. 

And Charlotte Moore-Lambert noted that a lot of people have to fight, and fight hard, for the privilege of getting homework.

Malala Yousafzai and other young women holding up signs supporting girls’ right to education.

Unfortunately, it’s hard to make any class that engaging. I try, but I have to be honest that I’m not that good. Few are.

Update, 23 February 2026: Oh, here’s something I hadn’t even thought of. Security. Mett Seybold on Bluesky wrote:

Students are going to voluntarily give their campus login info to this thing(.) This thing that I, who fancy myself a pretty top-notch portfolio sleuth, have been researching for two hours & can tell you next to nothing about who made it, who funds it, what other companies it integrates with. 

Caveat emptor.

Update, 24 February 2026: Dr. Vanessa writes that the same people behind the agent above also launched an agent for instructors that... logs into Canvas and does all the marking.

And thus, the circle is complete. 

Update, 25 February 2026: I suppose in the interests of (raises hands to make sarcastic air quotes) ✌️✌️ fairness ✌️✌️, I should point to an interview with the creator of this agent. I still don’t want to dignify it with names.

I do find it interesting that the creator of this agent did not finish a university degree, and pretty clearly views what goes on in university as “labour” that nobody wants to do.

And here’s another interview with the homework agent guy

Jacquelyn Gill observed:

Since yesterday, the… website underwent some rebranding. The tagline changed from "(It) does the busywork so you don't have to," to "(It) is the personal tutor every student deserves." In the FAQ, "How does (it) do all this?" became "How does (it) help me learn?"


External links

America will sacrifice anything for the college experience 

The shadow scholar 

Malala dot org

19 February 2026

Case dismissed against academic publishing

I’m not going to pretend I’m a legal scholar, but I did call this one from the beginning.

Over a year ago, a lawsuit was filed against several academic publishers, alleging they conspired to exploit scientists.

The case didn’t even make it to a verdict. It was dismissed back a few weeks ago, according to Stat News reporter Jonathan Wosen.

I’m a little surprised that I am only stumbling across this now. I do realize that academics have a lot of more dire events happening every day, but given the complaints about publishing practices, I would have thought to see a little more commentary on academic social media.

External links

Federal judge dismisses lawsuit against academic publishers over unpaid peer review 

Related posts

Scholarly publishers sued 

16 February 2026

Limitations to scientific progress

As a comparative biologist, I appreciated Nanthia Suthana’s new opinion piece about how neuroscience research is relatively divided by what model species researchers are working on.

I am interested in an assumption underlying Suthana’s thesis:

As a result, neuroscience’s primary limitation today is not a lack of data or tools, but persistent fragmentation across model systems, recording modalities and analytic traditions.

This made me wonder how fields assess their progress. Judging from conference attendance and journals, neuroscience is a phenomenally healthy field of research. Yet it is a field that somehow seems to think that, darn it, we should be further along.

How do research disciplines measure their own progress? To put it another way, if we were able to successfully remove a suggested limitation, what would we know?

If we got the “cross species dialogue” that Suthana thinks neuroscience needs, what would be the thing we would learn? All I can gather from the article is that we could better “refine or revise” our theories. But I’m not sure which theories those are, or what discoveries we might expect.

I realize that it might seem a big ask to get a preview of what discoveries we might make if we did more comparative biology in neuroscience. Unexpected lucky findings are the norm in every field of science. But in some fields, it is very clear about what certain limitations are, and what could be learned if those obstacles were removed.

Astronomers knew for decades that they would be able to see deeper into space if they have a space-based telescope.

Particle physicists knew for decades that they could test for the presence of the Higgs boson if they had particle accelerators that operated at higher energy levels.

Paleontologists knew for decades that major evolutionary events, like vertebrates living on land full time, should be in rocks of a particular age.

But there are a lot of fields that are just not like that. Neuroscience might be one of them. Or maybe it isn’t like that yet.

Related posts 

Nominees for the Newton of neuroscience 

External links 

Neuroscience has a species problem 

 

06 February 2026

Politics and pendulums

A lot of people, and organizations, assuming the status quo is immutable. So they talk like politics is all swings and roundabouts. Some days you’re up, some days you’re down. That the “pendulum will swing back.”

But societies aren’t pendulums governed by physical laws.

They ignore that many societies have undergone irreversible changes. Often sudden, sometime calamitous.

Organizations, in particular, get so accustomed to “normal times” that they have no crisis mode. 

I am reminded of this because Science magazine – who I have often criticized for underplaying threats to American science – is at it again. This week’s editorial argues that the real wins for science are all quiet backroom deals, and that loud protests don’t get stuff done. I’d analyze it more, but luckily, Joshua Weitz already did that.