It’s the end of an interesting trip to Google IO in California, and this final day delivered plenty of content just like previous days. It began with two sessions on Artificial Intelligence. Over the last few years, many of you will have seen AI programs playing old-school video games such as Pong (or Bat and Ball if you are British). It’s almost magical to watch the computer play, getting more and more accomplished with each attempt. You can even watch it evolve tactics, as it learns to send the ball off at a steep angle to make it more difficult for the opponent to return.
Prior to the sessions at Google IO, I’d assumed the code to implement this was (a) very difficult, and (b) lengthy. A few years ago when the feat was first achieved this was definitely the case, but Google has some excellent tools available nowadays to make it much easier to develop AI solutions. The AI model to play Pong is literally about 10 lines of code. You obviously need to know what you’re doing to design the model, but it’s impressively little typing to implement. To make further progress in developing AI programs, it will be necessary to brush up on my old A-level Maths, and possible my first-year degree Maths; but the tools at Google have most certainly brought it within reach to add AI functionality to applications. If you have ~2 days spare, Google provide a useful looking crash course on Machine Learning.
Incidentally, many of the demos at Google IO were given using Colab (https://colab.research.google.com/notebooks/welcome.ipynb) – a browser-based development environment that even works on Chromebooks. It provides an excellent playground for quick software experiments.
Now, change of topic completely, another noteworthy session on the last day was about Flutter. Flutter is Google’s cross-platform tool for developing iOS and Android apps. It’s directly up against things like React Native. I was in two minds about Flutter. As you would expect, it is very nicely integrated into Google’s IDE and tools, but it didn’t seem to offer anything dramatically better than the competition – it suffered from “and me” IMHO. Furthermore, it doesn’t cover web, whereas React does and allows you to share more common code (theoretically). If performance is critical for a specific project, then Flutter may be the winner, but it would need to be a particularly extreme case. I am sure that Green Custard will use Flutter at some point, you learn by doing, and we need to be knowledgeable on all relevant technologies – that’s a big value proposition for us to our customers. But it will probably need to raise its game to displace React for the majority of projects.
That’s the end of Google IO for this year. We’ll be sure to enter the draw for next year, and who knows, our number may come up?