My new boss did an interview with me [crosspost]

figureLand insight, my new employer, is a startup that provides property developers with an interactive mapping tool to explore various land and property datasets.  I’m on the software team, working as a full stack developer and machine learning evangelist.

The following post, by CEO Jonny, was originally posted here, on the Land Insight blog.

We thought this would be an interesting interview as a lot has been said recently about Big Data, Machine Learning and Artificial Intelligence, but our company is one of the few that are actually implementing these new and powerful technologies. The soon to be “Dr” Manson has come straight from his cognitive neuroscience PhD to extract the insight from our data.

What is neuroscience?

It’s a term coined in the 1970s and can mean lots of different things. I prefix myself with “cognitive” which means I’m interested in the algorithmic rather than the molecular or psychological level of the brain. As in, the story of how neurons work together to store memories and drive behaviour, rather than how they function at a very low level.

What are some examples of things you’ve worked on?

I was interested in the hippocampus, which is a kind of internal map and GPS for the brain. In particular I wanted to know whether the system was able to do complex spatial calculations, a bit like the navigate tool in Google Maps. To test this I programmed a robotic ball to chase a rat around a room, while I recorded from the rat’s brain (yes, really!).

That puts a funny picture in my mind!  What got you interested in the brain?

I studied maths and physics as an undergrad but had always had an interest in the brain, which I remember from back when I was 15 listening to the Reith lectures, by Ramachandran. He talked about how people had different parts of the brain hard wired differently than others, so they can see colours when they hear orchestras play, or they can spot anomalies in patterns by combining information that didn’t deserve to be there. I thought it was fascinating about how the brain could do that and give people a super power.

How can neuroscience help you in the technology sector?

In neuroscience I encountered lots of messy multidimensional data and I had to come up with ways to display it. If you display data in the right way people can find patterns, so I’m very well placed to be taking in large datasets and making them easily available to people who can make decisions on that data. Its co-incidental that I worked on projects with a spatial element.

Can computers find patterns in datasets that are not seen by humans?

It’s not fair to say they can’t because they probably could, but there tends to be a question  first and you end up finding results that you didn’t expect. For example, if you ask ‘how long does it take an area, on average, to approve a planning application?’ we might randomly find a spatial pattern, like coastal towns approve planning applications more quickly. Then digging deeper there might be a population bias in there, such as an age or wealth demographic.

And this is good information as it can affect investment decisions.

Absolutely. And there can be a tonne of patterns and the more information you can possibly make available the more questions you can ask and the more interesting the answers will be.

What are the technology advances that are going to enable the insight from this data to emerge?

Well, at the moment I’m applying a deep learning algorithm to large amounts of text.  The output of the algorithm is a syntax tree – the kind of thing that an English language researcher might draw. Using this, the computer is able to “understand” the content of the text and pull out useful bits of information, which can be stored in a database for humans to explore.  Ultimately the plan is to feed this database back into a machine learning algorithm to find patterns, and insights, hidden within the data on a larger scale.

How can this be applied to Land Insight?

Once you’ve built all this data which is useful for people making decisions you can begin to automate the insights people want from it. So if you are looking to find a development site and you have all this information about a plot of land, in principle you could teach the machine to find the kind of sites people are looking for to provide them with more leads. In the meanwhile, we’re use it to scale up improving accuracy and display data more easily for people.

Are we really in the next industrial revolution?

I think over the next few decades we’ll see a number of industries automated. If cars can be automated in the next few years then public transport, taxi drivers, infrastructure will see a significant percentage of people are out of a job. And the same could be true for basic construction or building. Why have a man put cement on bricks and then bricks on bricks if a robot could do it? Of course every industry has its own quirks which will need human input but the bottom line is if a person can do it a machine can do it, it’s just a question of how much time do you want to spend building that machine.

So AI is coming?

Oh yes. By the time we have full AI it will no longer be a surprise. Over time we will get used to hearing ever more news stories of machines supposedly passing the Turing test (the ability for a robot to trick a human into thinking it’s also a human) – it’s just a question of how intelligent the robot has to pretend to be. In fact, I watched a robotics competition last year where the robots had to demonstrate how they would help in an emergency scenario (like the nuclear disaster in Fukushima)  – the first time you see a robot get out of a golf buggy you are like “oh my god, that robot was sitting in a golf buggy, and now it’s standing up outside it”, but after watching another couple of robots do it you are no longer impressed or surprised. In the space of a few minutes you completely change your view of what a robot can and should be doing. Over the next years, we’ll just adjust what we expect: “ of course you can speak to your phone and have it give you a good answer or have it give you a medical diagnostic. Of course you can have your house built for you by robots”.

But will a robot ever say “no, I don’t want to do that, I want to do this instead!”?

I have no idea. There’s no reason why robots have to replicate our version of cognition. But it might be that certain aspects of our cognition are important and that gives them their own will. It might just say “no, I don’t like you, I’m just gonna go and sit in the corner and read the paper”

What appealed to you about Land Insight?

I looked at a lot of different startups in the last few months and most of them didn’t seem like they were doing anything exciting but Land Insight is solving a real problem, and to an extent has already part solved it.

Being able to solve a real world problem by making data easier to access and easier to explore really appealed to me and is a highly valuable service to developers and the wider population.

And what are your ambitions outside of work?

I have been learning Russian so have been organising with some of our freelancers to do practise sessions after work.

Come on, don’t forget your weightlifting ambitions

Haha, I haven’t done anything for a couple of weeks. I’m up to 120 kgs now though.

 

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s