VentureBeat

Nintendogs for the DS was a brilliant game. Its premise — pick a puppy and nurture it to adulthood — played on the digital pet theme that has driven high sales from Tamagotchi to Pikachu, and let you teach your dogs fancy tricks and skills that would improve over time, almost like it was learning.

But it wasn’t. A Nintendog’s behaviors were scripted, its personality restricted to a handful of breeds. On the other hand, Novamente, a company that aims eventually to develop a super-smart artificial intelligence (AI), says that its current technology will make such limitations remnants of the past.

The company claims that its virtual dogs, seeded only with basic skills and the desire to get food, water and rewards from their owners, will try almost anything to achieve these ends, including spontaneous behavior that was not written directly into the code. And dogs are just the beginning.

It might start out sitting and rolling over, but if that didn’t earn it some kibbles or praise, it would start to experiment. It might suddenly attack another avatar, bark at virtual squirrels, chase its tail or climb into its owner’s lap. If it watched its owner’s avatar hiding food behind a couch whenever another avatar was around, it might start hiding its food. If the owner jumped up and down every time they greeted the dog, the dog would follow suit, as long as the feedback was positive.

The primary limits on these virtual dogs’ possible actions, says Novamente founder Ben Goertzel, would come from the range of available animations and interactions with other virtual characters and objects, not from Novamente’s AI. But, as animations are not Novamente’s forte, this part will be a challenge. Its focus is on creating and enhancing a virtual brain.

Let’s build machines that can think.

There are a few companies trying to create human-level intelligence in a machine, also known as artificial general intelligence, or AGI. The one with the most buzz is Numenta. Founded by Jeff Hawkins, who also founded Handspring and Palm, Numenta is trying to build an AGI modeled on the human neocortex, the part of the brain responsible for spatial intelligence, perception, motor functions and language. Its goal is to take this model and apply it to problems that require sophisticated pattern recognition, like machine vision, robotics and data mining and analysis.

Novamente’s Goertzel says that a human-level AGI could be developed in a matter of years and that he has designed a framework in which to develop it. Built over the last seven years, the Novamente Cognition Engine uses a modular architecture designed to let specific specialized parts of the engine — some for memory, some for learning, some for language — guide every process but simultaneously distribute each one across the entire system.

See this diagram below.

While Goertzel says that he has developed a theoretical framework and a system architecture that can enable a super-smart machine, he acknowledges that the hugely ambitious project still has a ways to go. Without substantial funding, Goertzel admits, Novamente will not be able to build the AI on its own.

The company has thus decided to make major parts of the Novamente Cognition Engine available as open source. This project, called OpenCog, received a small boost in January when the Singularity Institute, to which PayPal founder Peter Thiel is a major donor, agreed to pay two engineers to work on the project full time for a year. In June, Google, itself, decided to sponsor 10 OpenCog interns (mostly PhD candidates) for the summer. And on Friday, Novamente launched OpenCog Prime, a wiki that documents Goertzel’s entire vision — the theory, the architecture, and documentation on nearly everything that has been built so far. Read the rest of this entry »

Before the advent of the written word, the story goes, humans had to either store all their memories in their own heads, or by oral tradition passed down through designated members of their tribes. With trade came notation of facts and figures, and later alphabets, books and libraries. With them came the modern brain, which treats recorded knowledge as an extension of itself.

Throughout these developments, previous generations have grumbled that each new advance leaves us worse off — take this month’s issue of the Atlantic, which includes a feature story titled “Is Google Making Us Stupid?” by Nicholas Carr. Yet Google’s search-and-retrieve functions are only the tip of the Internet iceberg, when it comes to memory. A whole new generation of efforts to move our memories online is in the works, and may represent one of the biggest upcoming movements in computing.

Pensieve, an IBM technology, is the latest project to unveil itself. The idea, being able to snap pictures of business cards and people with your cell phone for later retrieval, sounds almost identical to Evernote, a company I reported on a month ago when it came out of private beta. That doesn’t mean IBM is copying; rather, that IBM is taking the most obvious tack first. Business cards (as well as receipts and other short, printed matter) are easy for image recognition software to read.

The problem right now is that the low-hanging fruit is fairly limited. Recording is easy — so easy that a Microsoft researcher has been doing it for nine years, saving photos, videos, web pages, and nearly everything else he interacts with.

Each of those capabilities is now duplicated for regular people. “Life casting” became a minor fad with Justin.tv last year, and more recently Qik’s had its public launch. Other companies like Kyte also offering video and picture feeds from your mobile phone, all of which can potential be saved. Emails have always been possible to save, although companies like Zimbra and Xobni have since added much more functionality, while Xoopit helps search through mail. For web pages, there are bookmarkers like Delicious, and upcoming services like Twine, whose private beta I’ve started using to save my web ramblings, although the service itself still needs plenty of work.

However, there’s still a lot missing. First, you need an integrated storage spot for all this material. Hard drives die, photo sharing services go down, email accounts are hacked. It’s likely that in the future web companies will exist that offer ironclad storage for all your data — meaning the complete, unedited record of your life. Storage services abound right now, but users will want something special for storing their lives.

Almost as important are editing services to narrow all the incoming data to points, which can be disseminated across Facebook feeds, weblogs and other public forums. If you really did record your whole day today, you’d have to spend a lot of time searching out the moments that mattered and tagging or annotating them for immediate use or later retrieval. The more automation exists, the more people will record parts of their lives.

Search and editing, in fact, are choke points that may stunt the growth of a memory industry. But then, there are trends that suggest otherwise. Image recognition, driven by advertising uses, is advancing rapidly under the care of companies like Blinkx and Viewdle. Voice recognition has stalled researchers for years, but companies like SoliCall and VoiceBox may yet offer a working solution.

Once software can recognize pictures, video and audio in addition to text, the work passes on to the growing ranks of semantic startups. Twine itself isn’t just a storage point for web pages; it’s attempting to add structure through automatic, intelligent tagging so that when you’re trying to find something you’ve saved, it’s easy. (A similar effort not yet out of stealth is called Qitera.)

Such startups will represent the first set of technologies that can truly help establish external stores of memory. Simulating short-term memory early startups like ReQall and Jott, both now available on the iPhone, already help with day-to-day reminders.

Our long-term memories are the tougher nut to crack, but there’s a wealth of opportunity in automated journals, work streams and research logs, not to mention data mining services that can help us manage our time better (RescueTime is an early example). And a true integrated service may be closer than it seems; take a look at Numenta, which is working on a “hierarchical temporal memory system (HTM) patterned after the human neocortex.”

And when all these things exist, what will happen to our memories? As the Atlantic article suggests, we may find that the net effect is to “scatter our attention and diffuse our concentration,” — or, as it argues in another part, we could spur a “golden age of intellectual discovery and universal wisdom.” The result may well depend on the quality of the efforts.

Some very smart people believe machines will take over — directing human affairs — sometime soon, perhaps within the next thirty years.

So much malarkey, you might think, until you consider the sorts of technologies being developed lately — some of which are starting to replicate human intelligence, one step at a time.

There’s Silicon Valley’s Numenta, the company started by Palm founder Jeff Hawkins, which is building a computer system that aims to think like the human brain. We’ve mentioned the company before, but Dean Takahashi, who starts writing for VentureBeat next week, provides an update about the fascinating company. Numenta’s software copies the human brain’s way of recognizing patterns with hierarchical memory (such as recognizing a dog shape; see image above and explanation here). Some 100 or so developers and eight companies are working to develop the software, and are applying it to several application areas, including visual object recognition and speaker voice identification.

It is the latter area, voice identification, that is seeing significant innovation — in part because of the booming mobile phone industry it serves. At the Mobile World Congress in Barcelona last week, another Silicon Valley company, Audience, emerged saying it has build a chip and software modeled that aims to function like the human ear in order to improve the sound quality of cellphones. It suppresses background noise by mimicking the human ear, which breaks down sounds into recognizable components of speech, and can thereby ignore other sources of sound. See the New York Times story about the Mountain View, Calif. company.

It was started by Lloyd Watts, a neuroscientist who studied with the physicist Carver Mead of Cal Tech. Paul Allen, a co-founder of Microsoft, along with several venture capital firms including New Enterprise Associates, have pumped in more than $22 million into the company, we’ve learned.

Forest Baskett, an investor at NEA, told VentureBeat that Audience blows away competing technologies. Incumbents include Forte Media, which uses technology that forms a beam in order to track your voice, but which can experience interference if background noise finds its way into the beam. There’s also Softmax, which uses multiple microphones, and assumes it can sort out which voice source to focus on, but can also breaks down in busy environments like airports, where you can have five to ten sound sources at any given time. Qualcomm recently bought Softmax.

Yesterday, another voice identification company Solicall, of Israel, emerged saying its sophisticated technology serves Windows Mobile, and beats existing sound identification technologies. SoliCall’s software focuses on the specific voice of the speaker, using a personalized filter that first learns your voice and afterwards, during any call, passes through only your voice and screens all other voices, noises and sounds. It’s based an algorithm for which the company has filed a patent.

These bite-sized developments, where technology starts to mimicking the human, aren’t doing it the centralized Frankenstein way, but from the edges, one step at a time.