Where are the Intelligent PDAs?

Hello Sammy and the PalmAddict crew,

Where are the Intelligent PDAs?

I have been thinking about the evolution of Personal Digital Assistants and the current state of their intelligence. Looking back to the original Apple Newton and the first Palm Pilot up to the current state of PDAs, they really haven't evolved that much and are still very "dumb" computers. While the term "Personal Digital Assistant" implies that these devices will actively help us organize our personal information, and to some extent there were promises they would, when it comes to Personal Information Manager (PIM) applications, today's PDAs are still basically Personal Digital Organizers, passive devices that simply store information that their owners enter manually. There are no intelligent features that learn our patterns, anticipate our needs, and carry out common tasks automatically. A good secretary or personal assistant does these things for their boss automatically or with very little direction after learning their boss's routine habits, schedule, contacts, and personal style. Sadly, our digital assistants aren't there yet.

When the Newton first came out, Apple commissioned a movie called the "Knowledge Navigator" (Quicktime Streaming Movie) that showcased their vision for an "intelligent" personal digital assistant. If you haven't seen the movie, I recommend watching it and then thinking about the current state of PDAs.

In the movie, the owner engages his PDA in conversation, and the Knowledge Navigator understands, provides information like an encyclopedia or resident expert on almost any subject, and carries out commands. While just a concept, the Knowledge Navigator is a true personal digital assistant, message center, and communications device using voice and video teleconferencing, not just an electronic version of a Franklin Day Planner.

Apple implemented some "intelligent" features into the MessagePad 2x00 series and eMate 300 running Newton OS 2.1 called the "Intelligent Assistance Architecture":

• Helps users complete repetitive tasks including communications, scheduling, finding, and reminding
• Contains smart defaults that anticipate your needs, saving you time in entering data
• Integrates data from different built-in applications
• Supports third-party extensions

You can see the Newton Assistant in action in this short clip. It's relatively simple but brilliant. Write "Meet Bob for lunch on Friday", the Newton flawlessly translates your handwriting to text based on the training you've provided. Tap the "Assistant" button and it recognizes that you'd like to schedule a meeting with Bob for the following Friday at 12:00PM (lunch time). A "slip" for 12:00PM on the next Friday with "Bob" as an attendee is opened automatically, all you have to do is confirm the details and it's done. The Assistant automatically looks up all contacts named "Bob", and if there is more than one, you select which Bob you're meeting. You can easily change any of the details before finalizing the entry.

I bought a mint MessagePad 2100 back in 2003 to check out the Newton's handwriting recognition and the Newton Assistant. I was immediately amazed. To this day, no other PDA has the "intelligence" of the Newton, and if you've ever owned or used a Newton with the Newton Intelligence Architecture, you know what I'm talking about. For the most part, PDA users still rely on manual text or data entry through rudimentary character recognition that is no where near the true, trainable, and very accurate Rosetta handwriting recognition in the later Newtons. No other PDA can anticipate the user's needs and help complete repetitive tasks like the Newton could almost 10 years ago. Add true speech recognition, and "real intelligence" to the lunch meeting with Bob example mentioned above, and it's easy to see that your PDA or smartphone could truly become a personal assistant along the lines of the Knowledge Navigator concept.

The main goal of Jeff Hawkins's research into human intelligence at the Redwood Neuroscience Institute is to understand human intelligence and then create intelligent machines with "real intelligence", not "artificial intelligence". I'm still waiting for this type of intelligence where my PDA learns my behavior, my handwriting, my voice, my routines, understands my handwritten and voice commands, anticipates my needs, and helps organize my life by carrying out common tasks automatically. It should also learn and understand contexts and relationships.

Combined with the social nature of our mobile gadgets (see "Social Machines" in this month's Technology Review Magazine) intelligent PDAs could learn to recognize who our friends, family members, and business contacts are, how we communicate and interact with them, and help us communicate, collaborate, and share in new and innovative ways. Just look at the increasing popularity of photo sharing services like Flickr and mobile social software services like Dodgeball (which Google recently acquired), and you can see that our computers and heldhelds are social devices. That's one of the main reasons why non-connected PDAs are "information islands", losing marketshare to connected devices like smartphones and PDAs with integrated Wifi and Bluetooth connectivity.

While Jeff Hawkins's comments in a recent interview about a secret third business at Palm may be referring to the LifeDrive and the "new" Mobile Manager category of devices, I'm hopeful that his work at the Redwood Neuroscience Institute will eventually result in PDAs with more "real intelligence". While this may be years away, I'd still like to see more intelligent features and capabilities incorporated into our devices. If Apple could mimic intelligence almost a decade ago, we should demand more innovations in our PDAs that are long overdue. Its time to put the "personal assistant" back into PDAs through the use of "intelligence".

- Brian (Massachusetts, USA)