Learn

This means that we are fully in development! Alfie functions well enough to support teachers, but there are still many things we want to improve. So, you can expect quite a few big and small changes in the coming period.

We need users to find out what we can do better. Only when many people with all kinds of different needs and devices use Alfie will we discover where things are going well and where they are going wrong.

Exactly! We're curious about how you use Alfie in your lessons, what types of lesson materials you create, and how you formulate learning objectives. We want to know which features you appreciate, what you don't understand, and what you downright dislike. And if you let us know, then we can make Alfie better!

You have a few options: the feedback button at the bottom left, the 'like' and 'dislike' buttons on the generated materials, and you can, of course, email us at support@openalfie.nl! With this, you have a significant influence on the future development of Alfie.

Definitely. We've done our best to fix the biggest errors, but Alfie will occasionally still crash, give nonsensical results, or just not do what you want. Let us know when that happens, and we'll try to fix it as quickly as possible!

That's pretty much the deal. We can't yet guarantee that Alfie will always work, and you help us by giving feedback. In return, you can use Alfie for free. Sounds good, doesn't it?

Please do! The more feedback, the better.

Oooh, so much! First of all, more materials: additional learning activities, creating projects or even whole series of lessons. More possibilities such as working with graphs and images. And, of course, ways to increase the quality of the materials, such as subject-specific prompts and a library with examples. We also want to improve the interface and give you more control at all stages of design. And much depends, of course, on the feedback you give. So for now, we have plenty to do :)

At some point, we will decide that Alfie is good and consistent enough for a version 1.0. Then we will also offer a paid version. What exactly will happen for free users we don't yet know exactly, but the goal is to still offer a free version.

Currently, we use OpenAI's language models. For creating teaching materials, we utilize gpt-4-turbo. This is a more accurate and reliable model than gpt-3.5, the model behind the free version of ChatGPT.

In short: a language model analyzes large amounts of text to learn how language is structured and what relationships exist between different concepts. Then, it uses this knowledge to generate new text based on the input. If you want to know more, CGP Grey makes a good attempt at explaining the process of training AI: How AIs like ChatGPT learn (English video, 9 min, Dutch subtitles available). The basics behind language models are explained by the Guardian: Chatbots Visual Explainer (in English). For the ultimate nerds among you, there's also this article: The Sorcery Behind GPT (in English).

No. The text generated by the AI depends on what it deems most likely based on the texts it has been trained on. It doesn't necessarily know what is 'true' and what isn't. Sometimes it confidently produces entirely fabricated stories — this is called hallucinating. However, the quality of AI has now reached a point where the output is correct in the vast majority of cases. This makes it a useful tool, but one that still needs to be checked.

The book Chatting with Napoleon by Barend Last is both well-founded and practically applicable. aivoordocenten.com also has quite a few good tips (both in Dutch).

Barend Last, Tom Naberink, and Marcel Mutsaarts regularly post nice (Dutch) articles on LinkedIn. Be sure to look them up!

Then I would book a workshop! You're definitely in good hands with the people mentioned earlier, but we also offer workshops. Quite good ones, if we may say so ourselves :) We don't only cover the workings of AI and how you can use it effectively, but also how education can change and has to change in a world where AI is a thing. Email us at support@openalfie.nl for more information.

Tokens are a measure of how much 'AI' you use. Each time you talk to Alfie or have Alfie create teaching materials, it costs tokens.

Under the hood, we use the artificial intelligence of OpenAI. Each time we call upon it, we have to pay. We convert this cost into Alfie tokens. We would like to eat tomorrow too, so we keep track of what you use and set a limit on it :-)

No, printing, downloading, and emailing do not cost any tokens. We handle these processes on our own server, and that's waaaaaay cheaper.

During the beta, you will get 200 Alfie tokens per month. If your tokens run out, you will get new ones on the first of the following month.

Click on the token symbol in the bottom left, and then you'll have your answer!

That's not a question. But hey, email us at support@openalfie.nl and we'll see what we can do ;-)

It could be that the total monthly ceiling has been reached. That means all users together have used so much AI that we are now broke. Hopefully, we won't get there, but if so, we'll send an email shortly with what you can expect.

We're not sure yet. What we do know is that at some point we will have to start charging to keep running. For now, we're thinking of a subscription where you get a significant amount of tokens, with the option to buy more for heavy users. Don't want a subscription? We also plan to give everyone a small amount of tokens after the beta, but we don't know how many yet.