Now that Facebook’s digital assistant is available in the wild, one can see how simplistic it is, indicating just how far Facebook still has to go to get a real grip on making its services more intelligent.
Facebook M has been in beta for over 18 months and has comprised a combination of automated responses and human interactions, where the vast majority of the tasks have been carried out by humans.
The problem with using humans for “Digital Life” services is that it is very expensive to scale the service for 2 billion users – especially when the service will be funded by advertising. This why Facebook is working as quickly as it can to develop its in-house expertise, and while it remains a laggard in AI, it has shown some progress.
For example, at its F8 developer conference in April, it showed some good progress on machine vision, enabling its apps to recognize the world they can see through the smartphone camera. It also made Facebook M available to US users and most recently in Spanish to users in Mexico and the US.
However, what has gone live is only a small part of the grand plans that were announced in 2015 which envisioned an always-on, all-knowing bot with which the user could do almost anything. Instead, Facebook M is limited to suggestions – referred to as “M suggestions” – which are contextually sensitive pop ups that appear in messenger when the user types messages. For example, “hello” (or “hola”) results in the suggestion of emoticons that are waving, while “tomorrow” can result in the suggestion of a link to the calendar to create an appointment.
The available functions are very limited, leading me to believe that each function has been manually programmed using statistical analysis – meaning that there is virtually no AI in the service that has launched.
Although the service is extremely limited at present, Facebook has created a placeholder to be upgraded when it’s ready, as well the possibility to generate some data that should help improve what is already there.
However, outside of mixed reality, the immediate applications for this in Facebook’s ecosystem remain quite limited. This reinforces my opinion that Facebook is way behind when it comes to AI, and that the biggest challenge it faces is to bring its AI in line with that of its main rivals.
The problem is that its rivals are starting to use AI to improve the depth, richness and utility of their services, potentially leaving Facebook behind. To keep up, Facebook currently throws humans at its AI-related problems (eg fake news and objectionable content), which is clearly not scalable. Unless the AI problem is fixed, Facebook will have to employ more and more humans, leaving its EBIT margins, valuation and competitiveness at risk.
Facebook has some time to address this problem, as its newer Digital Life services of gaming and media consumption have scope to keep revenue growth going in the medium term.
This is why I like Facebook’s investment potential but I am waiting for the short-term fall in revenue growth (see here) to be priced in before pulling the trigger.
This article was originally published on RadioFreeMobile