Ben Garside

Ben Garside

Ben is a Learning Manager for the Raspberry Pi Foundation making teaching resources for educators. When he's not spending time tinkering with his Raspberry Pi, he likes spending time with his family

Activity

  • Keep checking Raspberry Pi as well as we're soon going to launch a set of lessons on AI that uses Machine Learning for Kids. Should be available to download by the end of March.

  • I have plenty of training data to provide that model!! :)

  • Classification could be used for this problem, but sometimes AI is not necessary. This could be solved (arguably much more effectively with less margin for error) using a simple IF statement

  • Good examples. What parts of social media applications are powered by AI do you think?

  • Hi Andrew, I'm really pleased that you've enjoyed our courses. It genuinely makes our day when we read things like that.

    With regards to next steps, it really depends on your stage of life and experience. That said, I asked a few people at work and below is a summary of their responses:

    1 - Depending on your location, it's worth looking at your local...

  • I thought it might be which is why I wanted you to share it!!

  • If you can, try and detach the space example from the decision tree.

    Let us know which part doesn't make sense and I'm here to help :)

  • Let us help Aida. Think about what you know about unsupervised learning. We provide the labels an algorithm to be created to recognise the patterns in the training data. In unsupervised learning, we don't provide the labels, we just provide the data. The unsupervised learning approach means that the patterns are found in the raw data. Patterns that we humans...

  • Sounds like a good structure

  • Dave..how could you?

  • Hi Ali, good to have you with us on the course. I hope that you enjoy it

  • Hi Anthony. Great to have you with us on the course. I hope that you enjoy it and it fills the gap left by less Netflix time!! Keep us posted with your comments as you progress through the course.

  • Yes if the rules are simple enough and the dataset is small. If it was a very large data set, it would take most humans far too long to work out the optimum splits.

  • You're absolutely right. The benefit of decision trees is their high level of explainability.

  • And it's reassuring to know that all of the above is possible isn't it? As with climate action, AI and ML can be real game changers if governments embrace it.

  • Hi Jasmin. Thanks for signing up. I really hope that this course continues to interest you. Keep commenting on the steps and let us know how you're getting on :)

  • It might be worth refreshing the browser or reuploading your scratch program. Let us know how you get on.

  • That's a shame. Did you try it in a different browser completely?

  • Welcome Naushin! Glad you're on the course. I hope you find the course useful. Keep us posted with your comments.

  • That's really interesting David and the first time we've had someone find an error!

    What I find interesting is that decision trees always give a prediction with 100% confidence. It makes sense as they follow a route and will always land on a leaf, but as you have pointed out, the prediction is not always accurate

  • That's great to hear Paul! Glad to have you on the course. I hope it gives you lots of ideas of things to add into your sessions going forward!

  • And that's really important to learn as well. The AI model worked as it should, but the application that uses the model is where the error was. Often people think as applications as being "AI" when actually the applications use an AI model. An important distinction I think. For example a smart speaker uses AI models, but not all of the functionality of a smart...

  • Really glad that you enjoyed it David. Looking forward to reading your comments on week 2.

  • Absolutely and machine vision is used for that purpose. Do you think it can be relied upon without a human expert?

  • @GlennHowlett Excellent to hear Glenn. @GlennHowlett and @DavidRichardson we're about to launch a series of 6 lessons on AI in partnership with Deepmind. You can read more about it here: https://www.raspberrypi.org/blog/experience-ai-deepmind-ai-education/ but you should see the free resources launch at the start of April.

  • Good example. I have seen photoshop recommend edits to your photos and that certainly is AI

  • Exactly and for ML models to give accurate predictions, they need lots of data, certainly more than we're reasonably able to manually type in, but as you might imagine, movie streaming services collect this type of data automatically from your viewing habit.s

  • Good recommendation, I will check this out.

  • Yes I wholeheartedly agree with your points Aida. Although I am pleased you got directed to this course. Well done AI!

  • Welcome to the course Arjen. Sounds as though you have quite a bit of experience in this area. I hope that you enjoy this introductory course!

  • Pleased that you're finding it interesting and thanks again for your contributions to the course! Looking forward to finding out how you find week 3.

  • You are correct in that people look different ages depending on a number of factors. Don't forget that you are placing human factors/perceptions on your thinking. If a machine is analysing pixels, we don't know what relationships are found in the data that make someone look the age they are. Would factors such as the time of day, makeup etc change the results?...

  • This is exactly right David. Great answer. There will always be bias in machine learning models. Once you're aware of that it helps you think clearly about the steps you can take to try and minimise it as much as possible.

  • Are talking from painful experience Tom?

  • @JerryYoungblood Glad that you got it working and using this for translating sign language is a great idea!

  • Great! Really pleased that you're enjoying it and thank you for your contributions to the comment sections!!

  • Exactly! The large bookmakers almost certainly use AI to make predictions about likely outcomes on sports fixtures

  • I enjoyed looking at that visualisation, thanks for sharing

  • Sometimes it can cause problems when your webcam is being used by another application. Perhaps make sure that the teachable machine browser tab is completely shut down before running your scratch program.

    Otherwise paste a link to your project and we'll take a look

  • There's a great quote from Dale Lane in the previous step where he mentions that the worlds best algorithms won't perform well if the dats isn't good quality.

    Programmers create the machine learning models, but you don't need to be a world class programmer to be the person responsible for training the model.

  • It's great that you're thinking carefully about the training data and the impact that has on the accuracy of your model.

  • Often these types of model farm lots of data to use to train the model. For example it might be from collecting data of viewing habits from a streaming service, or it might be collecting data from an online clickbate type quiz

  • It's likely that whilst incorporated into the same app, the food delivery apps use more than one machine learning model. In your example, one for the navigation and another for the food preparation time

  • Great answer!

  • Post us a link to your project Ian, let's see if either me or your fellow course participants might be able to spot any bugs

  • Glad that you enjoyed it Ian. Hope that you find week 2 equally as interesting

  • Hi Richard. Glad you could join us. I hope that this course gives you lots of ideas and inspiration to bring into the classroom.

  • @AidaJakob Yes that's a good example and I'd say it was knowledge organisation

  • Great examples. ChatGPT certainly seems to have caught a lot of attention. Hopefully this course will give you an understanding of how they work and realise that whilst they are undoubtably impressive, there's no magic.

  • Hi Ian. Welcome to the course. I hope that you enjoy it.

  • Depending on the functionality it could use many of the above categories. Look at knowledge organisation, what functionality in your watch could use that?

  • Well done Aida! Really pleased you came back to it and had success. Please reach out if you find anything else daunting as we're here to help :)

  • I'm really pleased that it's sparked both of your interests. You're both asking the right questions about what happens under the hood. We start to go into that during week 2 of the course, but it's week 3 where we go into that in a lot more detail.

    Thank you for your contributions this week. Looking forward to more discussions during week 2!

  • I'm glad that you enjoyed it Andrew! You can download it by clicking "file" and then "save to your computer". It will only work using the Scratch software which you can download here: https://scratch.mit.edu/download

    You imported your model into Scratch, but there are fun tutorials on how to develop websites and use JavaScript to import your model into a...

  • That's correct Marusa!

  • Good example

  • That's correct :)

  • Duolingo certainly does use AI. You might find this link interesting: https://blog.duolingo.com/how-duolingo-experts-work-with-ai/

  • This is great Andrew and you've touched on the fundamental difference between rule based systems (where you have to teach the machine every single command you wanted to use) and AI models which you will explore further in this course. The AI models are able to recognise lots of different voices and commands as they have trained using lots of data to work out...

  • Great example!

  • That's correct. It's a machine learning model that Deepmind called "AlphaGo". It's really worth checking out what other projects Deepmind are developing. AlphaFold in particular is extremely exciting for the future of humanity.

    https://www.deepmind.com/research/highlighted-research/alphafold

  • Welcome Ali. Good luck with the course.

  • Hi there David. Glad you could join us on the course. I hope that you enjoy it!

  • Hi Jamey, glad to have you with us on the course. I hope that this course helps spark your curiosity even further

  • @AidaJakob How close is your hand to the camera? For example is camera capturing lots of background data for the model to process?

  • Why do you think that is Kesley? Is there anything you could do to improve the accuracy of your model?

  • @AndrewScott Thanks for sharing Andrew

  • Hi Terefe. Glad that you could join us on the course. I hope that it's useful to you and you come away with ideas on how you could teach this topic to your students in Ethiopia.

  • Welcome to the course Ahmed. I hope that you enjoy it and please keep the comments coming as you progress through the course.

  • That's really interesting. Thanks for sharing. I think this is a really good example of how to engage young people in AI and ML and allow them to think about its potential. Do you have a link to any video you think would be good for other course participants to watch?

  • Hi Andrew! Really pleased that you have enjoyed the courses, I hope that this one also lives up to your experiences so far. Looking forward to reading your comments as you progress through the course.

  • I think you're right. It shines a light on the potential for future workplace changes.

  • What do you think makes it "intelligent"?

  • Absolutely. Good example

  • GPT is a really interesting development. Mainly in the fact that it's been made available for free (for now). What impacts do you think this will have our lives?

  • Welcome David. I hope that you enjoy the course.

  • Good to have you with us Gopal. I hope that you enjoy the course.

  • Hi Stuart, welcome to the course. Glad to have you with us. ChatGPT has caused a lot of disturbance in the education world. How do you think it will change the education world?

  • Hi Kenneth, glad to have you with us on the course. I hope that you enjoy it and come away with ideas for the classroom!

  • Hi Aida. Welcome to the course. You don't need to be a tech expert to get involved and have fun on this course. I'm looking forward to hearing how you get along. Please keep us updated by commenting on the steps as you go along.

  • Hi Rita. I hope this course helps you get started with your AI and ML journey. Good to have you with us

  • Hi Jerry, good to have you with us on the course. Good to see that you've been exposed to the world of Raspberry Pi as well. Keep us posted with how you're getting along with the course.

  • Hi David. Welcome to the course. I hope that you enjoy it and get out of it what you need.

  • Welcome to the course everyone!! My name is Ben and I will be facilitating this course.

    We will be around from the 3rd of January till the 27th of January (2023) to answer questions, offer help and guidance and join in on discussions.

    This is such an exciting and developing area of computing and I am really excited to see how you all get on taking your...

  • Hi Leo, really pleased to hear that you have the big book of pedagogy!

    The intention was that the pedagogy applies to ages 5-18. As this course is aimed at Secondary level you will be able to see how the pedagogy can be applied to these age groups.

    I'd be interested to hear your thoughts are you progress through the course.

  • Welcome everyone! I'm Ben and I'll be facilitating this course from today (8th December) until 23rd December.

    I think programming pedagogy is such an exciting area of development in Computer Science education. I hope that this course gives you new ideas and tools to use the in the classroom, or even that it gives you confidence that you were doing a lot of...

  • Glad that you could join us Georgina. I hope that you find the course useful. Reach out if you have any questions

  • Great to have you with us. I hope that you enjoy the course :)

  • Great to have you with us. I hope that you enjoy the course

  • Hi Marwa. A lot of this course includes Python, but stick with it as the pedagogy applies to whichever language you use, including Scratch.

  • Great to have you with us on the course. I hope that you enjoy it.

  • That's really interesting. Is there much focus on programming skills, or is it more the use of applications?

  • Hi Oghale, good to have you with us on the course. Looking forward to reading your thoughts and comments as you work your way through the course. Good luck and I hope that you enjoy it.

  • That's really good to hear and thanks for sharing. I'm going to look up Mote to find out more about it!

  • Good to have you on the course Diagro. Where about are you joining us from and what is your motivation for joining the course?

  • Hi Mikhail, great to have you on the course with us and I hope that you enjoy it. Looking forward to hearing your thoughts and a little bit more about your context.

  • Welcome everyone! I'm Ben and I'll be facilitating this course from today (5th September) until 30th September.

    I think programming pedagogy is such an exciting area of development in Computer Science education. I hope that this course gives you new ideas and tools to use the in the classroom, or even that it gives you the confidence that you were doing a...

  • Hi everyone and welcome to the course. I'm Ben and I'm a Learning Manager at the Raspberry Pi Foundation. I will be facilitating this course over the next few weeks. I will be around to answer questions and offer feedback from today (the 5th September) until 30th September.

    I've been part of the team that developed the Teach Computing Curriculum here in...

  • @DebbieWinchester There are risks associated with both, but the argument is that the cloud servers are more likely to be more secure than your computer due to the amount of money and effort they will put into security. Safety is likely to be key part to their business models.

    I agree with what you're saying, but could someone access your data if they...

  • Welcome Danyal. I hope that you enjoy the course!