A visit to IBM’s 18th-century Hursley mansion base
As someone who doesn’t drive, I simply cannot wait for the day when driverless cars are omnipresent and trusted enough not to require a driving licence to operate. Yes sir, my 70s are going to be sweet.
For now, though, when IBM invites you to take a look at their latest innovations at their Hursley base near Winchester, it’s a heady mix of Tube, trains and taxi. But it’s worth it when you get there: an odd juxtaposition of cutting-edge technology research that is helping shape the world, all housed in the grounds of an 18th-century mansion. Bridging the past and present is IBM’s very own museum of hardware past, which is so mesmerising that it warranted its own page.
And of course the Grade II listed building itself is gradually being pulled into the modern day with Internet of Things connectivity measuring the temperature, pressure, light, humidity and sound levels of each room – a project done by graduate placement students in their spare time, with the eventual aim of full automation. As if any more proof were needed that IBM had truly left its mark on the old place.
There was no big announcement to write up on my visit: just a steady stream of cool things demonstrated by enthusiasts keen to share their handywork. In the spirit of that, here’s a taste of what I saw.
Watson joins the Starship Enterprise
Against all the odds, I found myself a victim of the Peter Principle and promoted captain of the Starship Enterprise for a quick demo of Star Trek Bridge Crew. It’s a VR game by Ubisoft where you and a bunch of friends take to the bridge of a Federation spaceship in the Star Trek universe – and because the captain gets to sit down, I’m pleased to report that, for the first time playing a VR game, I didn’t end up walking into a wall in front of everyone.
But what does this have to do with Watson? Well, as of this week, a patch will make the single-player game much more immersive when your friends are unavailable – or more likely haven’t plonked down £679 on a HTC Vive. Using Watson’s speech-to-text engine, you can now issue 160 or so commands to AI crewmates using the power of your voice. So yes, you do get to say “engage!” and “beam them aboard” just like the child in you always wanted.
More impressively, Watson understands context a bit more than a standard voice-to-text engine, so you’re able to freestyle a little bit. I’m sure I felt my AI crew’s eyes roll into the back of their skulls when I asked them to “whack up” an incoming transmission “onscreen”, but they did it without question. That’s right – if Watson thinks I’m the captain, so do you, my AI peons.
I had a beard when I was a baby
My first passport was registered as a fresh-faced 17-year-old, which meant that in the twilight of its acceptability, customs tended to do an elaborate double take. Equally, if a criminal has been on the run for two decades, it’s unlikely they’ll look exactly the same as their last mugshot, although this is something I have precious little experience of.
IBM’s Watson has been chewing through hundreds of thousands of photographs of people as they age, so now it can age you while you wait. Of course, I tried this:
The best judge of how fair the estimate of me as a youngster would of course be my Mum, so I sought her comment right away.
Nonetheless, it’s important to point out that the photography software never asked for my age (33), and yet that’s in the area that looks the closest to me now, if you can see past the face dots.
Said face dots were, of course, applied by the software. If you happen to be wearing sunglasses when the photo is taken, then your resultant mugshots could happily audition for some truly terrifying horror films.
Publicly available data can really make Watson shine
Another side project showed the potential of Watson to venture into new areas, using readily available public data. Given that this project was put together in a matter of hours, it’s unnerving how impressive it appeared.
TfL travel data has bus routes for London, there’s plenty of detailed mapping software out there, and Watson can learn what heavy traffic looks like through live camera feeds. Put these three things together and you’ve got a system that can help you tweak your route based on real-world information.
Sure, in practice, this may not be any more useful than our current solutions, but the really interesting thing here was how easy it was to train Watson to learn an entirely new discipline. You don’t have to use Watson in isolation – giving it access to publicly available data can really make it shine.
Encryption that can’t be read… even by those working on it
IBM’s security team is working on the practical uses of fully homomorphic encryption (FHE). For those not in the know, this is a type of encryption in which data can be sent and interpreted by a third party that never finds out exactly what it’s assisting with. Why would you want to do that? Well, IBM had a practical demo of the tech in action.
In this demo, a fictional corrupt nation state (Kish) is trying to deal with internal terror threats. It is co-operating with the US and UK embassies. One of the many cameras in the city picks up a known terrorist in the north part of the city, and needs to quickly establish if his known associates have also been spotted in other areas by running an encrypted search through the other embassies’ camera software. Even in these diplomatically trying times, the US is trusted enough to send through the data to be processed via end-to-end encryption – which is to say it is sent through encrypted in such a way that can be unlocked at the other end.
But with leaks and corruption in the Kish government, allowing their embassy to see the suspect’s identity would blow the whole operation wide open. On the demo, I’m shown the data transferred to the Kish camera network, which decodes the whole software without ever knowing what it’s doing, and the known associates – who look a lot like stock images of celebrity sportspeople – are sent back to the US embassy.
A HoloLens makes building maintenance look amazing
I hadn’t had the pleasure of using the HoloLens before, and it’s one of those things that you don’t really fully understand until you try it. A demo waterpipe unit was set up that could control liquid pressure, and it automatically flagged a problem with the IoT sensors in the building. But that was just the beginning: the HoloLens had been programmed to put the pipes of the building in an overlay, so looking around the room I could see exactly where the fictional problem was, as well as ID tags to know which spare parts would be needed to fix it. Fortunately, I wasn’t actually required to do the maintainence – me and DIY really don’t mix.
If I somehow were responsible for maintenance, however, I’d definitely want one of these:
That’s a connected hard hat, which will sound a remote alarm should it detect excessive smoke, a single hard blow to the head, or even repeated softer blows – a talent demonstrated to me with the liberal use of a cricket bat. No wonder the mannequin looked so moody.
The sheer diversity of projects on display was pretty impressive. IBM may not hog the limelight in the way it did for much of the 20th century, but it’s undoubtedly still here, and it underpins an awful lot of the tech expected to power the 21st.