Google i/o – The meh factor.

Reply to this post

A rather subdued affair.

• Android P offered up such marginal improvements that Google was forced to rely on previews of long-term work in progress to generate any meaningful wow factor.
• As has become customary, Google ignored the really important issues with its Android operating system and instead rolled-out a series of marginal improvement that included:
  • First Google Assistant will soon include the ability to pose follow-up questions and requests without having to say “Hey Google” every time.
  • This will be a welcome addition and shows some good AI progress as short-term memory is difficult and well-known problem in digital assistants.
  • This shows Google’s steady progress in the process to make the assistant easier to use and converse with and will further distance Google from its rivals.
  • The assistant will also now support multiple requests but how well this will work outside of the carefully scripted demonstrations remains to be seen.
  • Google has also fine tuned the assistant to work better on devices that also have screens for the most used scenarios such as YouTube, enquires and recipes.
  • Smart Home controls will also now surface on the relevant screen when invoked via the assistant in a nice user experience touch that is certain to please, but doesn’t really add any real value or functionality.
  • Second, Google Duplex was relied upon to generate some wow.
  • This is a new ability that follows-on from making the assistant more conversational where Google Assistant will make phone calls to businesses to book tables or services.
  • Many small businesses do not have online booking systems and where a telephone call is required.
  • This is a function that is carried out by human executive assistants making it a good showcase for Google’s AI.
  • What Google was really trying to demonstrate here was its ability to create an AI capable of having a conversation with a human where the human can’t tell the difference.
  • In computer science, this is known as the Turing Test and represents a huge step forward in AI if achieved.
  • However, under demonstration conditions there are too many unknowns.
  • For example: How scripted was the conversation? Did the human know? How much more widely could this be applied as the use case demonstrated was extremely narrow.
  • I suspect that Google focused just on the use case of booking tables and services and taught it everything it could about that use case.
  • This is how most digital assistants are programmed.
  • I very much doubt that this version of the assistant would be able to pass the Turing Test in any other use case scenario.
  • Hence, while Google Duplex is a great demonstration of where Google is headed, I suspect that it will not yet perform nearly as well in the real world.
  • This is why it is not being released and remains very much a work in progress.
  • Third, Android P: which has intelligence, simplicity and digital wellbeing at its heart.
  • In practice this means that Google has added AI to some aspects of the OS to allow it to perform better.
  • This includes adaptive battery which resulted in a 30% reduction in CPU wake-ups from learning what the user is likely to do in a certain scenario or at a certain time of the day.
  • Android P will also include new functionality that will allow developers to surface portions of their apps anywhere in the user experience as well as better volume and screen rotation controls.
  • Google has also made some upgrades to the home button to make it more intuitive and functional, but these will only work on pure Google Android devices.
  • Hence, the vast majority of devices that are sold with Android P will not have this feature rendering it moot.
  • Android P will also include functionality such as the dashboard which shows usage, unlocks etc.
  • The idea is to help the user manage device usage to help prevent excessive usage and all of the wellbeing problems that can result.
  • Fourth, Maps: which also sported upgrades using AI as well as including the camera as part of the experience.
  • Using AI in maps will result in the curation and suggestion of places and locations that the user cares about as well as the matching of places that the AI predicts the user will like based on his usage.
  • Also new is a technology that Nokia released years ago which is using the camera to superimpose directions and places on top of what the user sees.
  • This makes walking directions much easier, positioning more accurate as it triangulates off buildings for which it can recognise and knows the exact location as well as offering points of interest in a different way.
  • However, this did not work well for Nokia and apart from the aid in positioning and walking directions, I don’t see this as a very exciting upgrade.
  • Fifth, Google Lens: where Google has upped the ante in using AI for object recognition.
  • Many competitors are also trying to do this but with this update, Google is brushing them easily to one side.
  • Google Lens will now recognise text from photographs and allow it to be copied and pasted elsewhere as well as being able to match objects to others of a similar style.
  • This will be used for making recommendations where the opportunity for monetisation is obvious.
• Google is increasingly putting its world leading AI at the front of everything that it does and as a result should be able to drive differentiation from it.
• However, once again the real problems of Android namely its endemic fragmentation, Google’s inability to update its own ecosystem devices remained largely unaddressed.
• Furthermore, the fact that it had to turn to a product with no current launch date to wow its fans is a clear indication that outside of AI, innovation in smartphones is rapidly plateauing.
• The cost to the user of these hopefully useful features is data, as without the user’s data, these new services will work at all and Google will have no way to earn a return on its investments.
• Google is innovating from the cloud to take the fight to Apple because its inability to fix the most fundamental of its problems prevents it from competing on the device.
• In that regard, Apple’s AI is simply no match for Google’s but when it comes to the quality of the apps and services of third parties that Apple offers, Google remains hopelessly outclassed.
• Hence, I think that there is an opportunity to really close the gap to Apple, but for as long as it declines to fix its problems it will have no prospect of success.
• In this environment of user increasing awareness of privacy and personal data usage, Apple is likely to outperform all of its ecosystem peers.
• However, for those based on advertising, Google remains untarnished by the current furore and is likely to fare the best.