Q&A with our AI Supremo, Jessica Cooper

In celebration of ‘International Women and Girls in Science’ day, we caught up with Jessica, our AI researcher, for a Q&A session on machine learning, artificial intelligence and why this is such an exciting field of science to be in.

Jessica is part of the Predict Mobile team and specializes in AI and all that this discipline demands for the business. She is also very good at Scrabble!

 

Q1: Can you explain what your role is and what it involves?

A1: I’m an AI researcher – my job is to figure out how to solve problems using artificial intelligence, either by adapting ideas other researchers have published, or by coming up with new ideas. I spend most of my time reading academic papers, thinking, sketching, taking notes, and writing lots of code.

It’s creative problem-solving basically. Machine learning and computer science are unusual in the sciences in that they’re really more about building new things than about discovering things that already exist.

 

Q2: What made you want to go into this type of work?

A2: I wanted to work in AI because I think it’s maybe the most important and interesting field in the world. Everything is going to be completely different in 50 years because of this technology.

I think the appeal of research rather than engineering is that if you’re doing it right, you’re always pushing the frontier, trying to find new ideas and improve the state of the art. Of course, this can be really frustrating because it can be hard to know ahead of time whether a new idea is actually going to work in practice – and quite a lot of the time, it just doesn’t! So you keep thinking and experimenting and hopefully eventually you find a solution to a problem that nobody else has, and it’s worth it.

 

Q3: What would your advice be to young girls considering a career in science?

A3: If you are very curious and a bit stubborn, you’re in for a very interesting career! I’ve always been of the opinion that it’s best just to think of yourself as not a girl or a woman in science, but just as a scientist. Go and be an excellent scientist and let your work speak for itself.

 

Q4: How would you explain machine learning to a person just dropped into 2021 from the 1980’s?

A4: I’d say clearly you’ve mastered time travel, why do you need me to explain machine learning?

In fact, lots of machine learning ideas have been around since the 50s, the only difference is now we’ve got the computational power to actually run them.

The basic idea behind most widely used machine learning is about using all this computational power that we’ve got to approximate a function between an input and an output. This could be, say, taking a random photo as an input, and some label – say ‘hotdog’ or ‘cat’– as the output. We don’t know how to write a function that checks photos to see if they contain hotdogs or cats or whatever ourselves – it’s too complicated – but if we have enough labelled images, we can use machine learning to ‘learn’ the function. So you start off with some random function, give it an image, and compare the output to the label that you were actually expecting – and then change the function a little bit so that the output gets closer to what you were expecting. And you keep doing that until it gives you the right answer most of the time. For Predict Mobile, I designed a bespoke neural network that learns from historical usage data to predict future usage, based on the same principle.

 

Q5: Do you have a personal philosophy in regard to machine learning and ethics?

A5: I do think researchers have a responsibility to consider the implications of their work, but I’m fortunate in that the kind of things I’m working on at the moment – immune cell segmentation, deep network explicability, and of course the time-series predictions for Predict – don’t throw up many ethical dilemmas aside from making sure that the data is anonymised.

Like any powerful technology, machine learning can be used for good or ill. But I definitely don’t think that fear of misuse should stop us from building powerful technologies, just that we need to be thoughtful about how we ensure they make life better for humans.

People in Effective Altruism have a lot of very smart things to say about AI ethics and safety, so if you’re interested in this question, I recommend checking them out.

 

Q6: What does the future look like with machine learning being a default position in the world?

A6: I hope we’re going to see massive advances in all kinds of areas. We already are, actually – look at stuff like AlphaFold, Tesla, the GPT models. Machine learning is really sexy right now so there’s a whole load of funding being pushed into it, and we’ve also got huge amounts of data that we never had before, so I think it’s going to change pretty much everyone’s lives. It’s the most exciting time to be alive. Except for COVID, which is rubbish obviously.

 

A big thank you to Jessica for taking the time to chat to us about what being an AI researcher involves and the exciting possibilities machine learning opens up. We see its positive impact first-hand at Predict Mobile – thanks to Jessica and the algorithm – helping customers to always optimise their costs using our AI powered Early Warning System. And this is just the start of the optimisation journey with Predict Mobile.

 

If you have any questions, or would like to find out more about how Predict Mobile can help you save time and money finding your perfect supplier with the optimum bespoke tariff, get in touch through hello@predictmobile.com for a chat.