fwiw, I mostly agree with you (ai training stinks of some kind of infringement), but legal precedent is not favouring copyright holders at least for now.
In Bartz v. Anthropic and Kadrey v. Meta "judges have now held that copying works to train LLMs is “transformative” under the fair use doctrine" [1]
i.e. no infrigement - bearing in mind this applies only in the US. The EU and the rest of the world are setting their own precedents.
Copyright can only be contested in the jurisdiction that the alleged infringement occurred, and so far it seems that fair use is holding up. I'm curious to watch how it all plays out.
It might end up similarly to Uber vs The World. They used their deep pockets to destabilise taxis globally and now that the law is catching up it doesn't matter any more - Uber already won.
> fwiw, I mostly agree with you (ai training stinks of some kind of infringement), but legal precedent is not favouring copyright holders at least for now.
I know. I am describing how it should be.
Copyright was designed in a time when concealing plagiarism was time-consuming. Now it's a cheap mechanical operation.
What I am afraid is that this is being decided by people who don't have enough technical undersanding and who might be swayed by everyone calling it "AI" and thinking there's some kind of intelligence behind it. After all, they call genMS images/sounds/videos "AI" too, which is obviously nonsense.
The root problem is that the model reproduces Indiana Jones instead of creating a new character. This contradicts the statement that the model "learns" and "creates" like a human artist and not merely copies; obviously a human artist would not plagiarize when asked to draw a character.
the model isn't the one infringing. It's the end user inputting the prompt.
The model itself is not a derivative work, in the same way that an artist and photoshop aren't a derivative work when they reproduce indiana jones's likeness.
That does not seem obvious at all. Fan art and referencing is a thing, and there are plenty of examples of AI creating characters that do not exist anywhere in the training dataset.
That's why I said it's an argument by induction. Where's the limit for it to be different? 10 images? 100? 10000? Where does it stop being copyright infringement and why? Many people have paid heavy fines for much less. I don't think that "a billion images is so unfathomable compared to just one million that it truly is a difference in kind" is a valid response
Fair points and I appreciate the heads up. I'll add those things asap.
>If you're inviting teachers to add information about their districts and their students, you MUST take your security, your supply chain, and your disclosures seriously.
Definitely not the case with SlideHero. There is no facility to add student names or any school related information. It's purely a slide deck and activity generator.
>enter PII about minor children, and then get fired and that data will be invisible to my department forever.
That would be very poor judgement on the teacher's part for sure. There are no prompts to enter student data at all in SlideHero, it's really not designed for that.
>Please, be careful.
100% agree, I'll add a note to remind users not to add any identifying information, although they'd almost have to be willfully doing it since thats not in line with the purpose of the app.
Appreciate the concern though and you make valid points, so I appreciate it.
>I've worked as a teacher and I think people severely underestimate the amount of time teachers have to prepare lessons.
Yes absolutely. I was spending a lot of time in ChatGPT for brainstorming about a year ago and that's where the idea for SlideHero came about.
>That said, I don't think I would use it to create an entire lesson.
Agreed, the purpose of SlideHero is not to "take over" a teachers planning and lesson delivery, I've designed it to be an value add, which is why I devoted a fair amount of effort to the additional activites that come with every presentation.
>If I were still teaching, I would definitely try it out for that.
I'll take that as high praise considering you're not a huge fan of LLMs :)
>I would consider making it easier for teachers to share what they've created with each other.
Yes for sure. That viral hook is soon to come. I have ideas for a marketplace where teachers can list their presentations and make some money selling them too, but I'm getting a little carried away now ... that's for further down the line.
> I'll take that as high praise considering you're not a huge fan of LLMs :)
Heh:-) I mean to say that I've not really encountered a convincing use case for them before now. From my point of view, your tool is perhaps the closest to a killer application for LLMs that I've seen.
If there's one occupation that requires a full-time assistant, it's the high-school teacher. But as we all know, that's an unobtainable luxury so this might be the perfect use for an LLM.
> Yes for sure. That viral hook is soon to come. I have ideas for a marketplace where teachers can list their presentations and make some money selling them too, but I'm getting a little carried away now ... that's for further down the line.
I wish you all the best in this. It's very inspiring.
>but already on first glance there was a wrong image
Yeah the images are going to be mostly match, but there is a "swap" button to choose more suitable images where the ai has picked poorly.
>and some pretty ugly use of language
Was this in your native language? I'm not sure how well ChaGPT does outside of English.
>But as a starting point, I can imagine this is a huge time saver for a teacher if they want to discuss a topic spontaneously, and only have 20 minutes to prepare.
Yes absolutely! This is the goal of SlideHero.
>Even before the rise of AI I see lots of low effort lesson materials being used
At the end of the day, it is still up to the teacher to create worthwhile resources, this was true before AI and is still true today :)
Indeed swapping the images was easy and intuitive, I should have added that. The "ugly language" was a bit unexpected, as in general the support for Dutch is surprisingly good. Maybe the combination of the prompt and the translation, and the mixed used of languages. Perhaps if all the prompts were translated to a language of choice (an the user would be prompted for a language) it would work better. But maybe you never even considered making a multi-lingual tool, and just out of the box it more or less already supports that. This would have been unthinkable just a few years ago, truly amazing!
How do you combine a (full time?) teaching job with building such a tool? It feels way more than some hobby project. Congrats on the release!
>Perhaps if all the prompts were translated to a language of choice
Yes you're probably right.
>But maybe you never even considered making a multi-lingual tool
I think there is a way to produce output that is in the desired language, but I honestly haven't looked too deeply into it. For now I am going to stay focused on English though.
>How do you combine a (full time?) teaching job with building such a tool? It feels way more than some hobby project
With many late nights and coffee, lots of coffee :)
>if you can't be arsed to write it, I can't be arsed to read it
I don't write my own textbooks either ;) The fact is that the slides are just an aid. Teaching is performance art. A good teacher uses slides as a handy scaffold, the magic happens in person, when the teacher *presents*
>A good presentation is INFINITELY better than a mediocre one.
of course, and perfect is the enemy of good. I know it would be nice for every lesson and every presentation to be absolute perfection, but that does slightly conflict with reality.
>creating a GOOD one requires a deep understanding of the matter
which teachers have
>, a cohesive plan and direction for the talk to go,
which teachers have spent decades perfecting
>and the ability to imbue your personality into the slides as well, so they're complimentary to the lesson.
Yes of course. I think there's a slight mismatch with regard to how you're perceiving the tool. The AI generated content is more of a starting point. As a teacher I'm constantly refining adding to and removing content from *all* the resources I collect. There is no one-size fits all. It's no different with SlideHero.
I'll give an example. I've taught introductory Computer Systems to year 9 students 2 times per year to around 4 classes a semester for about 20 years. That's probably close to 200 times I've deliverd that course. I can teach the content in my sleep. I still find it useful to generate a slideshow because a) it's 99% faster for me to review the content than it is to write it by hand, b) there are areas like networking that I just don't enjoy so much and GPT does a better job than I do here and c) interestingly gpt will surface new ideas and perspectives I hadn't considered before!
Remember that the slides are fully editable. There is little expectation that a teacher will hit "generate" and present exactly what the ai produces. In the exact same way as I'll find a good YouTube video, but won't play it all. I'll skip around to the parts that are relevant to my particular class and sometimes even to a particular cohort.
>I know teachers are swamped with too much work and not enough time to do it in, but this feels like the classic Technologists curse
Believe me, AI in the classroom is a *massive* force multiplier. Good teachers can be great with AI.
>I mean, what if the answer is fighting to get more teachers with more time available to prepare their classroom materials
Perfect, but I got a class of year 9 history students tomorrow and I'm teaching a new topic that I'm not familiar with... a little help from ai is going to make my lesson *better* not worse, not least of all becuase I have more time to devote to other areas of prep.
> the magic happens in person, when the teacher presents
Yeah, I see that repeated often - usually by teachers - but that is definitely not my experience. Magic happens when I read and do the exercises. Teachers are a distraction unless they are of an exceptional level (think 3Blue1Brown).
Modern online education made this very clear for me. Sure, many online courses have teachers "presenting" and it is indeed nice to have a Nobel-prize-winning world-famous professor explain things to you in a novel way. No doubt about it. But 99,9999% of teachers I encounter in the real world are not anywhere near that level and I'm better off reading and doing exercising and, maybe, occasionally asking a question but I don't care for the "performance art" part of it one bit. I am literally doing my master this way and it completely beats anything I did in "real school" which consisted of distraction stacked on distraction on yet more distraction.
Your mileage, of course, will vary depending on your level of neural divergence and topic of study. Liberal arts like philosophy lean (slightly) heavier on interaction than, say, CS, but even in those cases your time is best spent reading a metric shit ton of books and processing them through introspective thought and exercises then watching performance art.
For the life of me I cannot phantom why teachers keep recreating the same stuff over and over again. Networking courses have been done thousands of time. Just settle on a nice basic course presented by a world-famous type and let the students do their thing without performance art/justifying your job.
Nothing comes close to just doing exercises and reading. I know I'm harping on the same topic here but teachers keep insisting we go to "classes" - sitting with 30+ noisy monkeys next to me - to watch them do their performance art and it is has become a major trigger for me for various reasons. A regular class with such a performance artist reminds me of people superficially interacting with an LLM and thinking they now master a topic. It's fun to watch and the teacher might mean well, but after an hour or so of this nothing has been gained and could be better spent studying and doing exercises. Have I mentioned exercises are important?
I know this is unpopular and I know I might be wired strangely compared to you. Also: I do not intend to come across as mean. All the teachers I met were very nice people and they do mean well. I'm just being straight.
Appreciate the feedback. I built it for k-12 educators and I've used it extensively in my own classroom. I can assure you that the tool is very useful and kids love the interactives a lot.
>most rudimentary understanding of a topic
This is exactly what a year 4 student needs!
It's easy to dismiss the content as obvious, but remember that a 9 year old is learning it for the first time!
> in depth lecture notes, source material, etc.?
Hehe I'm not sure a year 6 student is ready for all that :)
> Hehe I don't think Timmy in year 6 history is ready for all that :)
That's exactly what we did where I grew up... Starting from grade 5.
In history class: Analysing contradictory sources and observing how history gets made. In physics: Doing our own experiments and deriving formulas from that. In politics: Debating and negotiating resolutions, UN-style. In Latin: reading (simple excerpts) from De Bello Gallico.
Please rememeber that a slideshow is just one tiny piece of the teaching puzzle and SlideHero is not trying to be everything for everyone.
All those things you mention are good, and necessary (adjusted for audience age of course) and a good teacher will add all that value too and that can and should happen outside of SlideHero.
My goal is to reduce teacher load by just enough that teachers see the value in paying me $7/month :)
fwiw, I mostly agree with you (ai training stinks of some kind of infringement), but legal precedent is not favouring copyright holders at least for now.
In Bartz v. Anthropic and Kadrey v. Meta "judges have now held that copying works to train LLMs is “transformative” under the fair use doctrine" [1]
i.e. no infrigement - bearing in mind this applies only in the US. The EU and the rest of the world are setting their own precedents.
Copyright can only be contested in the jurisdiction that the alleged infringement occurred, and so far it seems that fair use is holding up. I'm curious to watch how it all plays out.
It might end up similarly to Uber vs The World. They used their deep pockets to destabilise taxis globally and now that the law is catching up it doesn't matter any more - Uber already won.
[1] https://www.ropesgray.com/en/insights/alerts/2025/07/a-tale-...