“This open-source AI voice assistant is challenging Siri and Alexa for market superiority” makes for a great headline. It is, unfortunately, completely false.
The AI voice assistant in question is Mycroft, and the so-called “challenging” of market heavyweights is beyond silly. Mycroft has 36 contributors, with minimal outside interest. A total of 104 developers have bothered to follow it on Github, and it has garnered fewer than 1,000 stars (a way for developers to register interest). Another project associated with data infrastructure, Kubernetes, meanwhile, has 1,191 contributors, 23,205 stars, 1,733 people watching it, and 8,163 forks. That is what “challenging X for market superiority” looks like.
In fact, the market is far too undeveloped for open source AI to stand much of a chance. For better or worse, fully-baked, AI-driven voice assistance is going to be proprietary for the foreseeable future.
What hegemony looks like
There’s a tendency to think that open source has won; to believe that the era of proprietary software is gone. However you may feel about software, this is clearly wrong. Open source does dominate certain areas of software, like data infrastructure. In other areas, like mobile, it controls a significant percentage of devices shipped (Android) while leaving a proprietary OS (iOS) to control nearly all of the profits generated in the market.
SEE Amazon Lex, managed AI services behind Alexa, now open to all AWS customers (TechRepublic)
In the world of AI, companies like Google and Amazon have started releasing more of their code under open source licenses. Because of this, Cloudera co-founder Mike Olson is almost certainly right to call out a trend toward openness in enterprise infrastructure:
[T]here’s been a stunning and irreversible trend in enterprise infrastructure. If you’re operating a data center, you’re almost certainly using an open source operating system, database, middleware and other plumbing. No dominant platform-level software infrastructure has emerged in the last ten years in closed-source, proprietary form.
It does not follow, however, that finished products like Alexa or Siri are at risk from amorphous blobs of open code like Mycroft. If anything, these proprietary services are thriving because they’re offering developers an easy on-ramp to voice-based AI services. Whether they’re “open” doesn’t really factor in. APIs, however, do.
Amazon’s Alexa has attracted significant interest–the platform now boasts over 7,000 “skills” (developer add-ons that perform a range of functions), a 7X jump since mid-2016. To further its reach, Amazon opened up the Alexa Voice Service in 2015, which gives developers the ability to embed Alexa’s tech into their products. Amazon expanded on this in 2017 with Amazon Lex, as it announced: “the same deep learning technologies that power Amazon Alexa are now available to any developer, enabling you to quickly and easily build sophisticated, natural language, conversational bots (‘chatbots’).”
Is it open source? Nope. It’s a managed service, which developers access through APIs.
Would making it open source help anyone? Not the target developer audience, as Amazon CTO Werner Vogels made clear when announcing Lex (as well as other AI services Polly and Rekognition) in November 2016:
[U]ntil now, very few developers have been able to build, deploy, and broadly scale applications with AI capabilities because doing so required specialized expertise (with Ph.D.s in ML and neural networks) and access to vast amounts of data. Effectively applying AI involves extensive manual effort to develop and tune many different types of machine learning and deep learning algorithms (e.g. automatic speech recognition, natural language understanding, image classification), collect and clean the training data, and train and tune the machine learning models. And this process must be repeated for every object, face, voice, and language feature in an application….[T]hree new Amazon AI services…eliminate all of this heavy lifting, making AI broadly accessible to all app developers by offering Amazon’s powerful and proven deep learning algorithms and technologies as fully managed services that any developer can access through an API call or a few clicks in the AWS Management Console.
Developers don’t necessarily want the underlying code to tinker with. As Vogels noted, most developers lack the expertise to delve into the code powering AI-driven voice assistants. It’s far better for them to “borrow” that expertise from Amazon (or Google, Apple, etc.) through an API call.
More than code
This could be where things are going generally, as DataStax executive Patrick McFadin related to me at OSCON, “Things in our industry are heading to a place where [software licensing] is irrelevant….Developers will use APIs and will be charged by the call or megabyte. The argument of what type of license will disappear as a result.” It’s certainly what developers expect in an early market like AI, when they’re hoping for services, not software.
Mycroft may end up being helpful to some developers, but it’s certainly not “challenging” the AI giants for anything. It’s similar to Diaspora, the open source alternative to Facebook. Remember that? It was a ridiculous notion that simply open sourcing the underlying code for a social network could actually deliver a network. The network, not the code, was the missing utility for Diaspora, just as Mycroft misses the point on the underlying intelligence that a company like Amazon or Apple can give developers.
It’s not just about code. It never is, really. The best open source projects are really about community. Code helps to encourage community, but it’s a small part of why communities like Kubernetes grow. In the case of Alexa, Amazon is winning the AI voice assistant battle because it’s making it easy for developers to become productive, fast. That’s easier to deliver, in this case, with an API call than with a Github repository.
Someone needs to tell Mycroft AI, Inc. CEO Joshua Montgomery. Montgomery has declared, “The seed money we’ve obtained allows us to position ourselves as the open technology in this space. Our goal is that when you think ‘open’ and ‘virtual assistant,’ you’ll think Mycroft.” Maybe he’ll get his wish. But no one cares about “open” in the way he wishes they would. Not for voice assistants. Not now. Maybe not ever.