Machine Learning changes how we approach Digital Assistants

facebook-m-assistant2

One of the two most interesting talks, for me of the Dublin Web Summit, have been two of the shortest. Both only lasted about 20 minutes, however both showcased the growing importance of Machine Learning and how it is being applied to change how we interact with technology and our devices.

Facebook M and Instagram’s Explore tool are two products, from the same company which are quietly looking to change how we are delivered content and how we use the web as it currently stands. Both use machine learning, combined with human intuition to develop platforms that learn and become genuinely useful to consumers as they grow and consume more information.

First, to look at Facebook M. This was announced earlier this year, and most certainly aims to take on Siri and Google Now as your go-to digital assistant. The key difference for M as it currently stands is that it doesn’t just rely on a computer supplying responses or results, these are tailored by humans who help teach the machine relevance and to better understand the query and how to respond.

The goal, is never to replace the human, but to make sure that Humans curate the machine to ensure that the personal touch remain, which can generate real usefulness for the end user. M has already been favourably reviewed, but it has been felt that Siri and certainly Google still have the edge, especially in relation to speed and utility.

Facebook believe however, that while Now and Siri are useful, they don’t provide value to you as a person. They don’t proactively engage you with reminders, or understand your behaviours or lifestyle as you ask more questions. They generally, also don’t answer intelligently. This is where Facebook looks to use Humans to add an intelligence to the information being processed and supplied.

The thinking behind this is that if you asked M to order you a Pizza, it will go an order you a pizza, rather than Now, or Siri giving you a list of Pizza delivery services. M will know the nearest one, it will have your details and it will take care of all the work, while you go about doing other things.

Facebook M goes beyond a digital assistant, it becomes a go to place for ordering, and purchasing. Facebook already has many of the world’s biggest brands and retailers seeing the benefit of advertising with them, every day. Instead of booking a flight on one app, then ordering dinner on another, before going to another to buy a travel guide, Facebook wants you to do all of this within Messenger, one app where all brands are present, which offers real utility.

Instagram take a similar approach to their explore feature, using machine learning. Extensive A/B testing using machine learning, combined with a human insight that adds common sense and intelligence to their Artificial Intelligence, have moved a feature that previously gave you images which may have been of no interest, to a section of the app which is curated by your own activity. This is something invaluable on an app as personal as Instagram.

Both of these things combined have looked at something which other services such as Siri and Now have missed. They aggregate huge amounts of data, but without much intelligence, providing information back but not much else. Facebook M looks to move the relationship beyond this to add a utility to a digital assistant as we grow to rely on them more.

The big change this helps bring is usability. More than likely, Siri and Now are not too far behind M, this will make these services really, actually useful as opposed to just handy, as they are now. Both, especially Google have brands and huge amounts of information at their disposal, already Google offers similar, but much more scattered services.

The big challenge will be how do they leave choice in the mix, people like curated content to a point, there are still some things machines, even human curated ones don’t understand. However, it’s an approach, and an insight that makes us shift our view on just how useful social networks, digital assistants and all of the data we supply actually are. Many companies have learned the hard way that walled gardens don’t pay off.

Wearables are the Next Big Brand Challenge.

Moto-360_Map-820x420

The average person on the street may not realise it, but everything they do is now being analysed, not just your online behaviour. The Dublin Web Summit is in full swing, and something that keeps cropping up is the growth of data directly taken from wearables and the consumer products we use, without a second thought – even your smartphone.

It’s no great surprise to anyone, that our activity is being analysed, your average person online is used to this from their web browsing. Google search ads, retargeting and Facebook ads are all now expected to be contextual to your online activity. You want to book a holiday, so after you’ve done a little research, you check out your Facebook feed to see an ad that has been generated by a price comparison site, or an airline, etc. This is something we now expect, even if we don’t always think about it.

The big change is that now, more than ever we are handing over data that goes beyond our search history. Many people now wear fitness trackers, often get to their ten thousand steps a day, people who don’t wear one will sometimes have an app that counts their steps anyway and log their activity. This information can, and will eventually be used by brands to create more specific and more targeted content with real relevance to the consumer.

To take the example of the fitness trackers, which are hugely popular. This could be a person who regularly reaches their step goal, or perhaps sleeps really poorly each night. It’s not too hard to imagine a brand-oriented platform which can take this data being generated and allow brands to promote on this. For the person who reaches their step count, it could be a healthy snack deal at a local supermarket, or for the insomniac, a local bed shop or blind fitters which could help them sleep a little more soundly.

This might seem a little Orwellian, but it’s not too far off where we are at the moment. Masses of anonymous consumer are supplied each day to brands looking to advertise. Currently, these are restricted to your search or browsing habits. While they are already very contextual, they don’t really understand you as a person, their data is limited to what you’ve chosen to share (which for most people, is quite a bit).

To put it in context, there are currently 17 billion connected devices in the world, and this is set to grow, massively with 52 billion by 2020, largely this growth will be driven by wearables as they become more mainstream. That’s a huge amount of information being pulled together.  Peggy Johnson, of Microsoft, speaking at the Web Summit said that as a company, they now view the person as the hub, with devices such as Band being built around that.

It might seem cynical at first, but to use wearables to help promote content, in a more context driven way would be more beneficial to everyone. Consumers would no longer get annoying ads which don’t really have any major benefit to their lives, and brands would no longer be able to deliver information much more effectively. The resulting content is much more bespoke and makes advertising do what it’s supposed to do; deliver insight to the consumer and to the brand.

However, with so many different kinds of contact points with a consumer, a brand should always remember that, it is at its heart a more intimate communication. Beyond a TV advert, or a Sponsored social post, adverts generated by wearables will be specific to someone’s actions and their lives, it needs a much more intimate approach than before.

This is where the challenge lies as new technologies reach the mainstream, combined with a growing expectation from people that the content they consume will grow to match it.