This week, we look at the remarkable performance of DeepMind’s algorithm, Intercom CEO’s perspective on chatbots and what precisely businesses are looking for as well and how New York has been trying to create a sustainable ride-hailing ecosystem.

Have a great week,

Hi Mum! Said Dad

DeepMind algorithm

Image: ITV

DeepMind algorithm set to be rolled out in the NHS in the next 3 years

It was announced earlier this year that DeepMind has established a partnership with the esteemed Moorfields Eye Hospital in London to analyse 3D retinal scans for diseases like glaucoma, diabetic retinopathy and age-related macular degeneration.

Things have moved on since then though. Initial findings published in the Nature Medicine journal recently stated that DeepMind’s algorithm was deemed to be better than eight retinal specialists at Moorfields over 997 patient scans. The Deepmind Algorithm had an error rate of 5.5% compared to between 6.7% and 24.1% for the eight specialists.

The other significant advantage of DeepMind’s innovation is the turnaround in results. The algorithm can analyse the scan immediately whereas patients would have to wait days for a human to review their scans.

Dr Keane, a consultant ophthalmologist at Moorfields has described the results as “absolutely jaw-dropping” with the tech planned to be subject to clinical trials before being rolled out in the NHS in the next 3 years.

There is a lot of noise about how AI is set to revolutionise healthcare but finally, it looks the British public are getting closer to experiencing cutting edge machine learning in their day to day experiences with the NHS.

By Jeevan Jayaprakash, Strategist


Image: The Verge

Chatbots: Now you see me, now you don’t

Over the years, chatbots have seen investment from tech giants such as Facebook and Microsoft. While some would argue that they have yet to revolutionise the digital landscape, companies such as Intercom disagree and believe that bots are every businesses’ future.

Today, bots are indeed everywhere. We may not always see them, but more automated messaging can be found on companies’ websites and apps than ever before. ComScore predicts that 85% per cent of customer interactions will be managed without a human by the year 2020.

However, Intercom CEO Eoghan McCabe, states that what many of his business customers are increasingly looking for is the consolidation of all their communication tools into one platform and a means to manage this as opposed to just having a chatbot (which is how Intercom started off). It is in this area where Intercom have doubled down and are seeing success. This is distinct to what Facebook is doing with Messenger and WhatsApp where only a front end is being provided. It seems Facebook is missing a trick.

Speaking of chatbots, my personal favourite is the highly impressive Mitsuku.

By Anastasia McLain, Strategist


Image: New York Times

How to manage supply and demand in a sharing economy?

Last week officials in New York approved a cap on the number of licences for ride-hailing cars operating for Uber and Lyft. London’s mayor Sadiq Khan is considering similar actions as a “necessary step” to clear up the city’s congested streets and improve working conditions.

The next step for NYC officials will be to announce the exact limit for ride-hailing cars. In the past, officials have spent years computing the number of yellow cab licenses New York City should hand out, and since 2009, the number of so-called “taxi medallions” have been capped at around 13,000. However, due to the dynamic nature of any big city, the number was set unsustainably low. The price of a taxi license skyrocketed to $1.3 million in 2013, and due to their relative scarcity, yellow cabs can be almost impossible to find outside Manhattan.

So, how should officials approach the capping this time around in order to generate a sustainable system for drivers, riders and the city?

A recent article by NYTimes argues for a data-driven approach. The ride-sharing market is essentially dictated by two variables: the mileage spent cruising for requests and the frequency of trip requests. Drivers ideally don’t want to spend hours driving an empty car, as it is time spent not making money. The same goes for the city, as this causes congestion!

The trade-off is that a low number of miles spent cruising means a high frequency of trip requests for drivers — leading to frustrated riders competing for constantly occupied cabs. The phenomenon is demonstrated in the graph above, created by the National Renewable Energy Lab using data for ride-sharing services in Austin, Texas.

From this type of graph, an optimum can be derived. For Austin, it lies at 3.4 trip requests per hour, which translates to having 30 drivers for every 100 trip requests. From here, it would be very easy to determine the amount of licenses that need issuing!

If Uber and Lyft agree to offer officials their real-time usage data, the license cap could regularly be adjusted to best fit the city’s needs. The question is whether the Silicon Valley giants are willing to cooperate with governments to create a better urban environment?

By Oliver Iyer, Strategist


Image: Pinterest

Cool thing of the week: Hotstepper

Tired of feeling lost on Google Maps? Stopping every five minutes to check for directions and looking like a dog chasing its tail as you try to figure out if you’re facing the right way? Forget the days of looking down and look forward!

HotStepper is a step into the future — your new slipper wearing best friend who will use his groovy dance moves to guide you to your destination. All you have to do is scan your surroundings, set a destination, and point your phone forward for this AR hero to show up. Don’t worry if you decide to stop though, HotStepper will wait for you to start walking before continuing to lead you on.

This use of augmented reality is going to change the lives of many directionally challenged people out there — myself very much included.

By Anastasia McLain, Strategist