Smartphones, Google Now, and Intelligent Agents

Smart mobile devices, such as smartphones and tablets, are quickly reaching the point where they will have the technology capable of becoming a truly Intelligent Agent (IA). The notion of intelligent agents has been around a while, with roots in the AI (artificial intelligence) and digital/ computing futurist communities.  The essence of the idea is an autonomous system that can take independent action on the part of users, based on observation of current context and what it learns from observing previous user choices and behaviors.  Or more colloquially, a digital personal concierge or personal assistant that can provide you what you want or need, when you need it.
  James Kendrick, in a post on ZDNet, suggests that we're quickly getting to the point where your smartphone can become your personal assistant and usher in a "new dawn of usefulness."  He spins a story of how smartphones as personal assistants can aid users throughout a typical day.  It's a nice example of "design fiction," where stories are told to illustrate the potential of technologies and applications.  (I still think that Corning's A Day Made of Glass is one of the best examples - and reflects many of the same uses and benefits, although displayed on a wide variety of nearby screens.  Still, you can think of the personal smartphone as the mobile interface that drives all the screens in the "Glass" world).
  Smartphones are personal mobile devices, and thus are in a good position to observe and record individual contexts, choices, and behaviors.  Current smartphones already combine anytime/anywhere access to the Internet and its wealth of information, geolocation tracking and apps, and the ability to monitor and record basic personal information and individual choices and behaviors that occur through the device. Recommendation systems are here and improving (if narrowly focused) - both Amazon and Netflix have strong recommendation engines that track user preferences and can autonomously search for and recommend alternatives that might be of interest.  Is this enough to usher in a new dawn of smartphones as intelligent agents?

  I'd argue that these factors alone aren't quite enough.  You also need to factor in the other side of the interactions - the wide variety of devices and systems that an intelligent agent will need to interact with.  The coffee machine in the kitchen, to start it brewing, monitoring and tracking public transportation, making reservations at restaurants, etc.  So far, only bits and pieces of digital systems are interconnected at the level envisioned in Intelligent Agent design fiction.
  Another issue is the question of the processing power, and storage capacity, needed for the better recommendation systems.  Current generation smartphones aren't up to the task, yet. For now, quality recommendation systems reside on their hosts systems, although their results are accessible on mobile devices.  In addition, even the best of existing recommendation systems are several steps below the level of artificial intelligence envisioned for Intelligent Agents.  The applications are extremely narrow, with minimal tracking of preferences and applied to fairly static inventory of choices.
  And then there's data service pricing.  Intelligent personal agents need to be active almost continuously - monitoring locations, searching for relevant information, updating personal preferences. They will generate significant data flows, particularly if hosted on the mobile devices.  As mobile service providers move from unlimited data pricing packages to data caps with punitive overage charges, users may not be able to (or want to) absorb the costs of operating Intelligent Agents.  And costs could skyrocket if the systems an agent queries for information charge for that access and use.
  Then there's the issue of mobile network capacity, when everyone's Intelligent Agents are driving those high levels of data flows simultaneously.  The London Olympics gave an unanticipated answer to the question of capacity, when the data flows from spectator smartphones along the cycling road course overwhelmed the official GPS tracking app designed to show where individual riders were during the race.  And for mobile IAs, that capacity needs to be wireless.  There is intense competition among the myriad users of radio spectrum, so it's not clear whether the mobile bandwidth needed will be forthcoming.
  The advent of Cloud computing and storage does provide a technological framework for hosting personal Intelligent Agents off-device - although serious issues have yet to be resolved.  Among them: the costs of required data storage and processing; privacy and data security concerns related to personal information and behaviors; offline back-up and restoration of agents in case of emergencies or failure of online systems; etc..
  Still, the technological and programming foundations for mobile-based Intelligent Agents exists today, although not at the level that experts expect IAs will need for smooth, efficient, and effective performance.  And you can see the beginnings of smart personal agents (not quite IA) emerge in prototype applications like Google Now.

  Google Now, at least in terms of what is envisioned, looks to be that kind of personal assistant - using smartphones to provide real-time location data to mesh with personal information (calendars, contacts, email) and public data feeds to create "cards" of relevant information that users can choose to access.  It looks to be using the Cloud to record location and Web browsing histories, and to perform the information processing needed to develop pre-packaged templates of personalized information that users can choose to access..  Bundled with the recent 4.1 upgrade for Android OS, it provides a glimpse (limited and imperfect) of what personal digital assistants and mobile Intelligent Agents can be.

Sources -  Smartphone of the not-too-distant future, your personal assistant,  ZDNet.com
Reading the tea leaves of Google Now, c/net.com

Slight editing for typos and clarity.