Photo by Joshua Woroniecki on Unsplash
This week we’re looking at AI and creativity. We are starting with some new ways in which AI is being applied, and looking at how humans are starting to interact with AI before moving into the education and learning spheres.
AI is already a huge part of our lives, and many of us may not even realise it or be conscious of the fact it exists. For some, AI is the stuff of movies and occasional news articles about beating humans at chess and Go, and for those that are more aware there is often a perception that AI is still strictly the domain of running big data sets and limited general intelligence.
However things are starting to change. As we will see, Artificial Intelligence is starting to enter the creative domains, traditionally recognised as being uniquely human and unlikely to be influenced by machines. This has implications for any creative and dynamic industry and form of work, including learning. So what are some of the trends that are happening, and what might it mean for us? Let’s find out!
We start with Hollywood, where movie scripts are starting to be approved by artificial intelligence using a platform named ScriptBook. From the website:
“Not only does ScriptBook reveal the financial probability of the screenplay in question, but it can also predict the MPAA rating, determine the emotions of each character, and separate the protagonists from the antagonists. Then ScriptBook selects the target audience, including gender and race. It can even inform you if your story passes the Bechdel test.”
ScriptBook has not and is not expected to replace human script writers and editors in the near future, but it has proven useful as a validation tool, and can help investors decide which screenplays might have a better chance of box office success.
Continuing in the Hollywood theme, a lot of movies now have a huge amount of high definition 3D computer generated graphics, and the technology is only getting more powerful. However the frame by frame rendering of these images is an expensive and time-consuming process for the artists, and AI has been identified as having huge potential in automating this process. The artists will continue to create and define the characters movements, but the routine rendering can be performed by AI.
In the creative domain, which was once considered a human-only space, Robots and AI are now creating digital art, with one piece selling at Christies for over $400,000 recently. To enter the Robot Art Competition and Exhibition, robots must use brush strokes, and it attracts those interested in robotics, image processing and machine learning. Contestants can enter original or ‘reinterpreted’ artwork, and it looks like that AI art is only the beginning.
When it comes to developing a general, human-like intelligence, AI has faced one big problem: catastrophic forgetting. Catastrophic forgetting is a challenge for AI because while it can recognise an object, it can’t transfer its learning to different contexts or perspectives, and the object is ‘forgotten’ and has to be relearned again, thereby preventing the development of human-style general intelligence. However, the code may be starting to be cracked. A paper was presented at a Human Level AI Conference recently that presents a way forward – an agent controlled by an AI algorithm. It can identify objects, predict how the object might look in other contexts, and ‘imagine’ what it might look like from multiple perspectives compared to other objects it has seen before. All of this helps the AI to ‘remember’, and despite the challenges of developing a general AI intelligence, it seems to be an important first step.
Are you ready for the first generation of app-based AI-powered coaches? Homecourt AI promises to revolutionise basketball training, and users film themselves making shots around the court, which are then analysed by the AI, measurements provided and recommendations made to the player. The Homecourt app identifies shot type, ball release time, release angle, player speed, leg angle and vertical jump height. Could we one day see PE teachers using similar technology for coaching assistance?
AI is becoming a part of our children’s lives, and this New York Times article looks at the development of the author’s young daughter’s relationship with Alexa. The author observes her daughter’s interactions, starting with saying ‘hello’ in the morning and asking simple questions and playing music to gradually handing decisions such as what to wear or play over to the AI. Research referenced in the article indicates that children instinctively trust AI such as Alexa, and they are too young and naive to realise the advertisement-driven model behind the recommendations it provides – in this case Alexa tried to sell the child the clothing it recommended. Concerned about what was happening and with limited checks and balances against the emerging power of AI, the author retired Alexa to the closet.
This article also looks at the interactions between children and AI, this time through the lens of AI-generated videos on Youtube with titles that don’t reflect their content. The AI generates content based on the indicators of what young audiences like, a lot of it doesn’t make any sense at all and that might be a problem, because “… an A.I has no goal, no ethos, no pathos, and no sense of wrong and right.”
It argues that with advertising becoming more powerful and harder to detect because of AI, parents have to work incredibly hard to teach their children to avoid websites and games that try to sell them things, increasing parental anxiety and leading to ‘drone parenting’, further limiting independence and increasing fragility in childhood. Evidence to support these concerns is limited, but it’s worth having the discussion.
What do you think would happen to the design of rooms and the expected flow of people in a school if given to a genetic algorithm?
This is an interesting study of how an AI can optimise floor layouts in a way that is designed for maximum optimisation without regard for convention or whether it can actually be constructed. The results are biological in appearance, similar to AI design in other industries, and may open up new fields and ways of thinking for designers to explore.
Coming to a high school or city near you, a site that provides camouflage from face detection algorithms. Style tips?
1. Avoid enhancers that amplify facial features such as cheek bones – they make you easier to identify.
2. Partially obscure the nose bridge area, as this is a key area for AI identification.
3. Obscure one of the eyes – the darkness and position of the eye is a key factor for recognition.
4. Don’t use masks (they may be illegal), but use makeup and covers to vary your features.
5. Obscure the shape of your head – another key identifier.
6. Make sure your face is asymmetrical – algorithms expect the two sides of a face to look similar.
Here’s an article looking broadly at how AI is impacting education. Some identified trends are:
- Automating administrative tasks
- Intelligent tutoring for ‘time on task’ knowledge and skill acquisition – see Carnegie Learning and Thinkster
- Identify individual needs using multiple data points, and identifying learning trends in real time and over a period of time
- Customised curriculum design
- Immersive technology such as AR and VR
- Platforms that power and integrate technology in the classroom – for example Clever products are in 50% of US schools
It will be interesting to see how this continues based on the growing power of the technology and how its being applied in the real world – it’s likely that the list above is only the very start.
To finish we thought it might be interesting to look back to The future of AI in education from 2016. The ‘grand challenges’ outlined in the article that AI should work to address are largely being tackled by current research and entrepreneurs, and it’s always worth looking back to check how things are progressing against the predictions that people make.
The research conducted and insights gained during the writing of this article have inspired the Indigo Schools Framework, the details of which can found in the Primer on our Resources Page. Send us an email at info@indigoschools.net or complete the form below if you’d like to learn more about how the Indigo Schools Framework can be successfully applied within your school. Also be sure to follow us on Facebook and Linkedin for our latest updates.