Two areas are of most interest to me. Firstly Android, which has ultimately become the software stack that delivers computing capability to the vast population of the world outside of the rich industrialised nations. Secondly, leadership in the development of the web itself, technical standards and rich software frameworks that comply with those standards and allow us to build interactive software that runs in the web browser on desktop and mobile phones. Google remains the most important commercial company in my research nexus, really the only one that I would sit through a keynote for. However I feel like I've gotten to know Google on a level sufficient to develop some reservations, perhaps even cynicism about their capabilities and motivations.
Ultimately, Google is a company filled with brilliant engineers. Engineers, so-called 'hard' science people, have a particular trait which I may elaborate on in another post. They are razor focused on their subject area and are either blind to or wilfully dismissive of areas outside of their specialisation, most notably the humanities. So when Google works on language, the form of AI that they explore just happens to be the form where it's not necessary account for anything in the humanities. Where the task can be boiled down to the universal approach of crafting a set of training materials for a generic AI model. It gets good results, but no one learned a single thing about the languages they translate. Google doesn't care. It's a technique that delivers a result. Well they didn't care, but it turns out that the current big thing is digital assistance. Where you ask for something and the computer needs to understand you. So we now come full circle and Pichai specifically mentions the importance of natural language processing.
My cynicism of Google can be summed up as follows; I saw them as an engineering-focused company dominated by the North American monoculture. Their simple and usable products belied an inherent overcomplexity of software engineering underneath. The Android platform is nothing short of disastrous from a developer's point of view, it's a complicated, fragmented and poorly documented mess which might come as a deep surprise to anyone familiar with how nice it looks from a user's point of view. Google's APIs more broadly I think are frequently the most annoying to use compared to any other commercial operator. Their large-scale documentation efforts seem like factories that general completely unhelpful materials, which itself has fostered an entire industry of people writing actual helpful documentation for Google-developed technologies.
Yet on the flip side, Google is changing. In the past you might have been hard pushed to believe a silicon valley giant would have a non-native English speaker CEO. Google has formalised their efforts on design, and design is very much a humanities driven area. Material Design is a well executed plan for universal design on desktop and mobiles, which rather than trade marking, Google provide free, and engineer excellent documentary resources and impressive software implementations, all for free. While Angular 2 is falling prey to engineering disease, the whole Angular ecology is one that remains immensely important for delivering interactive software on the web. Google recognised that the web can be an alternative to native platform development. So while they made Android, they are taking leadership in the advancement of so-called progressive web apps (PWAs), fundamentally addressing the question of why we need a native app in first place.
PWAs are the hot sauce for me right now. They enable a new powerful capability where we can, for example, send a link to a participant with a low-end Android phone. They open the link on their phone (remember, often the phone is the only computing device they have) and an 'app' appears which invites them to perform a set of tasks, all without being asked to install an app. In the 'app' ecology, this is considered to be a smooth 'onboarding' processing, and ultimately I've come to realise that it's a key factor if we are to realise any form of linguistic crowdsourcing.
Google is obviously driven by commercial interests, and this keynote shows this more than ever. After Pichai left the stage, the product roadshow began, busting out the usual array of nauseating silicon valley marketing-speak, where 'experience' was a high frequency lexical item. To be fair, it was never as bad as the horror of people like Microsoft of Apple.
Ominously, though, Google seem to have cottoned on to the fact that their cutting edge AI and digital assistance service would be best commercialised exclusive into their own range of hardware devices, ala Apple. I don't yet know how serious that is, partly because Google's digital assistant stuff is completely irrelevant outside of the American monoculture right now, and also that no one I know will be buying the high priced but otherwise unremarkable Pixel phones.
Fortunately it doesn't really matter. Google is a huge company and it's to be expected that entire genres of products and services just aren't relevant to us. Hopefully there will remain strong areas of crossover, like in Android, progressive web apps, Angular web frameworks, material design, and cloud solutions like Firebase (which I'm a huge fan of). In a way, I suppose we should be thankful that one company is doing so much that is helpful. If there's something I could wish for it's that other large technology companies were also doing at least one thing each which was an important contribution in the general overall theme of making technology more easily accessible and authorable by the ROW.