Artificial intelligence is transforming our world. From medical robots, to algorithms which combat climate change, right through to AIs that can help to ease world hunger, rapidly developing technology is finding answers to long-standing questions. It is teaching us to ask new questions which had never occurred to us before.
So, what are the hottest AI projects of 2020? Here is our top seven:
1. AI finds 50 new planets using old data
One of the great things about AI, is that it can help you derive value even from older data. Since the “data explosion” of the mid-2000s, many organisations have routinely collected more data than they were able to process.
The current generation of AIs, running on powerful modern computers, can finally take this data and process it at speed to extract new insights. It’s hard to imagine a better example of this phenomenon than the work of a team of UK computer scientists who built an AI to analyse historical data from NASA.
The British team trained an AI to analyse unprocessed data from the Hubble Space Telescope. Using training data that contained the characteristics of confirmed planets outside the solar system (exoplanets), the team trained their AI to distinguish a real planet from a false positive. The AI went on to find fifty new planets.
2. An AI wrote a blog post
Writers everywhere beware! The San Francisco company OpenAI has created a language prediction model, GPT-3, which uses natural language processing (NLP) designed to create texts that read as if they have been written by a human.
Argentine blockchain expert Manuel Araoz has taken the experiment beyond theory by training a deep-learning AI to use GTP-3 to write a blog post about itself. According to Araoz, it took less than ten attempts for the algorithm to successfully write like a human.
3. This robot wants to save the bees
Worldwide, bee populations have been in decline for most of this decade. And that isn’t just a problem for bees, it’s a problem for all life on Earth. Bees are one of nature’s most prolific pollinators, without them many flowering plants could not reproduce.
Katharina Schmidt, a long-time beekeeper and a computing science student in Germany, created apic.ai to try and find out what is happening to cause bee decline. The AI uses cameras to monitor the activity of hives enrolled in the project. It looks for patterns in things such as bee return rates, whether bees have successfully found pollen and so on, to see if it can find patterns which can help explain why bees are having such a hard time.
4. An AI that predicts whether you’re likely to have COVID-19
In the absence of enough tests, it’s been challenging for many countries to say with any uncertainty who has COVID-19. This has made self-isolation and assessing the progress of the virus more difficult than it should have been.
Researchers in London and Boston responded to this problem by creating an AI which uses symptom data from an app to predict the likelihood of a person having COVID-19, with no test required. The researchers found that their algorithm could predict whether someone was infected with an 80% accuracy rate.
5. A neural network that finds whales in the ocean depths
In 2018, researchers at the US National Oceanic and Atmospheric Administration approached Google and asked it for help. The administration had 180,000 hours of undersea audio recordings. It needed a way to “listen” to them all and find whale song in among the random underwater noises.
Google created an AI and trained it on audio with verified whale recordings. In 2020, this AI is now fully functioning and is helping the Canadian Department of Fisheries and Oceans to track the movements of endangered Orca whales in real time.
6. AI recreates ancient archaeological artefacts
Understanding the significance of ancient artefacts unearthed by archaeologists can be difficult at the best of times. When those artefacts are found in fragments, often with pieces missing, it’s even trickier. Researchers from around the world decided it was time to enlist help from AI. In 2019, Google Deepmind researcher Yannis Assael published a paper describing how he had used an AI to infer and restore missing parts of the text in Ancient Greek manuscripts.
Find out more about what you can do with an Artificial Intelligence online MSc at the University of Bath by requesting information and speaking to our online admissions team.