In this fourth part of our talent spotlight series, we speak with Dr. Joel Pitt, one of SingularityNET’s prominent AI Researchers, who has been instrumental in ensuring the latest computer vision and machine learning algorithms are available on SingularityNET. Working out of his native Wellington & Wairarapa, New Zealand, Joel embraces the vast nature around him and finds that it helps him keep up with the rapidly advancing field of AI.
Hi Joel! Thank you for taking some time to do this interview. Every AI researcher at SingularityNET has a fascinating background, and it’s always fun to learn more about them.
Let’s start with a topic that is in the news these days: creating art through AI. What are your views on that?
People question if it’s art or a just mechanical process when you use an algorithm to create an art piece. However, I feel it’s more important to ask what the artist is trying to say, and how it makes people feel rather than asking how it was made. Ultimately, I think that art should create a sense of wonder and joy in people or otherwise invoke questions about what it means to be human. If the art does that, it’s somewhat irrelevant as to how it is made. Unless the process is specifically part of the statement the artist is trying to make.
I’m excited about what people are doing with GANs for generating images as well as novel musical instruments, but ultimately there has to be something deeper to it for it to be an interesting artwork.
I take it that you have an interest in music. Have you experimented around with creating some music of your own?
I was a DJ many years ago and as such I’m into a wide variety of electronic music — amusingly people have also debated whether these computer-generated sounds should be considered legitimate music. But now it’s rare that music does not contain some software-manipulated sound in the composition.
So what activities take up your time now?
Lately, I’ve been spending more time on sports like long-distance running and mountain biking. Cardio work is known to be great for stimulating the brain and BDNF (brain-derived neurotrophic factor) production. Keeping active is important to maintain cognitive performance, and I can use all the help I can get in trying to keep up with the fast pace of development in machine learning and AI. Plus it keeps me sane.
It’s nice to get outside and whenever I feel mentally exhausted, running helps me feel refreshed and clears my mind. In that sense, it is a somewhat meditative experience — my awareness of time changes. Sometimes I even forget that I am running and suddenly discover I’ve travelled much further than I remember experiencing.
I think long distance running does that, especially once a person gets into the rhythm. Do you run daily?
I train 1–2 times a week, as my runs are long enough that my body needs a couple of days recovery to avoid injury. I’m preparing for a big event at the beginning of December called The Goat Adventure Run which involves running around one of the big mountains in New Zealand, in Tongariro National Park. This consists of lots of trail running, going up and down hills and crossing ice-cold rivers created by snow-melt.
I love the name of that challenge.
I was reading your educational history, and it said that you attended the University of Canterbury, where you convinced them to let you do the equivalent of a Bioinformatics Honours degree (half molecular biology and half computer science, with a pinch of high-level maths). What motivated you to do that?
Stepping back to my high school years, I had some idea on what I wanted to do. I read a couple of books which gave me a strong direction on what to study. One, in particular, was “Visions: How Science Will Revolutionize the 21st Century” by Michio Kaku. The book focuses on important questions like where humanity is going, what is going on with technology and what skills would be necessary in 15 years (at least as predicted from the 1990s!). So I tried to set myself up to be able to do something that would be both interesting and in demand. And two of these things were machine learning and computational biology. I was already able to program computers, but I found biology fascinating too. So I tried to hedge my bets.
I find AI and machine learning interesting because it lets us reason by analogy about the brain and consciousness. I think we in machine learning take a lot of inspiration from understanding the physiology and biochemical processes behind neural signalling in the brain, the evolutionary dynamics going on within species, and the complex dynamics between species at the ecosystem level. Even though neural network algorithms are very simplified models of what the brain does, I think we will continue to find inspiration from the systems that already exist in nature.
For your PhD, I remember reading that it involved Ecology. How did you go from Biology and AI to Ecology?
The foray into ecology came later. I took a detour to ecology and simulating ecosystems as I really wanted to do a PhD and was looking for funding as I couldn’t justify the cost of a PhD otherwise. I found an open position to do a PhD in a machine learning group at Lincoln University. They were focussed on ecology and bioprotection, doing things like predicting associations of species with different habitats and countries. This allowed me to do applied machine-learning work with ecologists while also doing some purely theoretical stuff.
While I like getting into the maths, I often enjoy the applied part of machine learning more now. I like trying to solve problems, figuring out how to make something work with the available tools under specific constraints. It might not always be the prettiest outcome, but the focus is on finding a workable solution to a real problem, and there is a lot of scope for being creative!
And what did you do after the PhD?
One of the projects that I worked on after my PhD, but building on that research, involved constructing ecological simulations for a government research institute in New Zealand. These simulations were predicting the spread of invasive species and weeds.
I used custom written GIS software, designed to run at the scale of an entire country — which, at the time, was a pretty large simulation. It was done in a probabilistic and stochastic way, running not one but thousands of simulations. In this way, you’d get more representative predictions that model the underlying uncertainty in the process.
New Zealand’s ecosystem is quite isolated and unique, so we want to protect our environment. If you visit New Zealand, you are prevented from bringing fruit or biological products in because they could threaten native species as well as our agriculture industry. In this case, if something did arrive, my software could help predict how an invasive species spreads. We can then sample and attempt to control the spread using traps, or herbicide to kill plants, and because this can be expensive, we need to be smart about where they go and when.
When I think of New Zealand, I think of stunning natural beauty and not a tech hub. How would you say is the startup and technological culture over there?
The tech scene here is obviously not as huge as in Singapore, Hong Kong, San Francisco, or other major centres but it’s still very vibrant. Having a lot of nature and wilderness attracts people to New Zealand. We also have a pretty decent work/life balance, so many talented people who have had enough of slowly working themselves to death choose to move here.
A good example is how the VFX for Lord of the Rings and Hobbit was done locally. Some of it was done with local New Zealanders, but international talent was also enticed to Wellington. A lot of cool things spawned out of that. Weta Digital helped bring a critical mass of people in specialized areas, some of whom then went on to do something on their own. One of those companies was my last employer, 8i.com.
There has also been an increase in smaller, self-funded startups. One startup I worked with a few years ago, called iwantmyname.com, is a domain registrar that has grown by giving people the best customer support possible. They realized that if you focus on the customer and make them feel valued, you get word-of-mouth marketing which can quickly make your business successful and profitable. It won’t ever become a behemoth like Airbnb or Uber, but it’s become successful through organic growth, and not only that the growth is sustainable and doesn’t require VC money.
Another thing that’s good about New Zealand is the ease of starting a business. You go online, register and pay — then you have a business. Over the years, there have been more opportunities in technology here and markedly more in the area of machine learning since I was in university. In the last year of my undergraduate studies, I signed up for an AI and machine learning course. Unfortunately, no one else did. So I begged the professor to do the course one on one. I didn’t know at the time if I made the right choice, and at times it’s been hard to find paid work that took advantage of my skills, but lately with the excitement around deep learning and AI I think it’s now paying off.
Considering that the demand for AI talent has skyrocketed over the past several years, I imagine that the AI course must be getting overbooked now.
As an AI researcher with so much valuable experience, I’m sure you must have already received countless offers from the large multinational corporations. What keeps you from joining them?
I did have opportunities with the likes of Google. But, I fell out of the process because my heart was never really into the idea of working for and being absorbed into a big corporation. Of course, they have good intentions to make the world a better place and make cool products. But still, there is always an incentive to create value for shareholders, and they have all these established practices in place, and it can be hard to know how or if I’ll really be making an impact.
I’m also concerned about how these large corporations influence human behaviour and how certain governments are using technology to control and surveil people’s lives. It sadly feels like some of the dystopian novels are manifesting into reality. Even though many of the people involved have good intentions, and some of us intentionally try to avoid it, it still happens.
I want to use technology to empower individuals. It’s always been the case that large organizations like governments and multinational corporations have far more power compared to individuals. Sometimes this power imbalance is important, say for dealing with criminals, but I’d still prefer that technology helps temper the power imbalance rather than amplify it.
Speaking of big corporations and lawyers, what are your views on patents in AI?
Early on, I was quite idealistic and thought that everything software should be open source. I still think it’s very valuable in terms of making it unnecessary to duplicate work and often seems like a better way for being transparent with software development. In some ways, the ethos behind open source movement ran counter to the idea of software patents, although the ideas are less in conflict these days. Regardless, it still seems absurd to me to be able to patent an invention that is purely algorithmic and be allowed to completely prevent others from using it.
For me it’s as strange as trying to patent a formula in mathematics, which is fundamental. If you start doing that, you prevent the progress of human knowledge. However I do recognise that people pour money into software innovation, so perhaps as an incentive, they should get a patent period for something like for 1–5 years instead of the default 20 years. I believe patents in technology should have a short lifespan. You should have a commercial advantage for your investment, but it should be reflective of the speed that the software world now moves at.
That makes me think of how SingularityNET plans to bridge that gap between algorithms and end users. So how did you first meet Dr. Ben Goertzel?
I met Dr. Ben Goertzel in the early 2000s through a now-defunct mailing list, called Shock Level 4. This was named after the psychological impact of sudden technological jumps. We started some conversations outside of the mailing list and I asked him for some career advice and what I should study. I was into open source, so he suggested I could help work on an open source project for a training environment for AI (a similar idea to the many of the reinforcement learning environments that now exist). I did that while juggling university. I was also given access to use Novamente, a predecessor of Opencog in my university projects. Then after graduating, I worked with Ben on a couple of paid projects.
Later, in 2011, I lived in Hong Kong for a year to work on the OpenCog project with Ben. Then I went back to New Zealand. The one year in Hong Kong was a great experience, but it made me appreciate living in New Zealand and the wide open spaces and natural environment that we’re privileged to have access to.
Joel, I can say with confidence that many readers (and myself) would also like to live in New Zealand and enjoy its natural beauty.
New Zealand is a great place to live, especially with the crazy and sometimes disturbing stuff going on around the world. It’s nice to be a little bit away from it all, while still being part of the global community. You get to enjoy lots of nature. And of course, I think Wellington is the best city in New Zealand. Obviously, it depends on what you are looking for, but it often places high on global rankings for one of the best places in the world live.
So you were back home, closer to nature, when did you decide to join SingularityNET?
Prior to SingularityNET, I worked with 8i.com on volumetric video. If you can imagine recording a person in 3D, then you can place that recording in VR or AR. They’d record people with a green screen and 40 cameras, reconstruct their recording in a volumetric video format that you can then stream and watch on any device — your phone, laptop, or AR headset.
That’s where I got into computer vision research. I was lucky to learn a lot from a great team.
Due to various reasons, I found myself being pulled away from being able to stay up to date with machine learning research. Ben was in Wellington for a few days, and we had lunch. He said that this exciting project is happening and I realized that at SingularityNET, I would have the freedom to learn about and implement a wide range of machine learning techniques.
So later on when Cassio and Ben invited me to join SingularityNET, I thought that a role more focused on machine learning research would work better for me. And here I am, trying my best to do cool stuff with SingularityNET.
Joel, it’s time for the ultimate question.
If I drop by Wellington, where would we go for pizza and which one would you recommend?
For a lazy night in, I would recommend a pizza delivery restaurant called Hell Pizza.
But for a proper sit-down meal, I would say Scopa on Cuba Street. It’s an Italian place that does pizza and pasta. They have handmade pizza — with a really good chilli oil. Since the pizza itself is so good, I would go with a standard Margherita pizza — with the chilli oil on top to make it absolutely perfect.
Joel, it’s been great talking to you! My resolve to visit New Zealand is at an all-time high now. Thanks once again for your time!
We started the talent spotlight series to showcase career paths that can lead to a meaningful and rewarding career in AI. If this episode intrigued your interest, we recommend that you read our interviews with the other AI researchers of the SingularityNET team: Dr. Deborah Duong and Dr. Anton Kolonin. You can also read about our community manager Tim Richmond, who shared with us his journey in the second part of this series. If you would like to join the SingularityNET team, please find our currently available jobs here.
In the future, we will interview more members of the SingularityNET team so that we can further highlight meaningful and rewarding careers in AI. If you have any further questions or would like to participate in the discussion about this interview, please visit our Community Forum.
For any additional information, please refer to our roadmaps and subscribe to our newsletter to stay informed about all of our developments.
Source: Deep Learning on Medium