Aug 16, 2018 | 9 min read

Insight Vector: The Coming Third Wave of AI – Technology That Truly Can Learn

We spoke with Peter Voss, founder of Aigo.ai, a software company developing natural language based intelligent agents that can be used for businesses and consumers. The company has developed an enterprise bot as well as a consumer bot that are designed to allow users to keep control of their own data, on contrast to the AI assistants like Alexa, Siri and Google, which collect data for their own purposes. Peter is a long-time trailblazer in Artificial General Intelligence, and he sees the next generation of AI incorporating true learning and adaptive capabilities that improve based on ongoing interactions with their users. 

 Peter_Voss

Peter Voss

Founder
Aigo.ai

orange-line.png

 

Can you share a bit of your background Peter? 

I started in engineering and got into software, went from a garage to IPO in a few years, and when I sold my interest in the company I focused on my next big project.  What struck me was how brittle software was – how can we build intelligent systems looking at how our intelligence differs from animals, researching and attending conferences? 

In 2001 I started my first AI company in Los Angeles focusing on various prototypes. I got together with Ben Goertzel (founder of Singularity.Net) and Shane Legg (one of the founders of Deep Mind) – the three of us came up with the term Artificial General Intelligence (AGI). This is not Narrow AI” or AI used for specific applications, the original dream was to create AGI.  

My R&D team developed an engine called Smartaction.ai which is still used in call centers. Customers like AAA and MGM Hotels, Hyundai and Carnival Cruises are using it. The company was doing well but I got frustrated because the applications were very narrow. Five years ago we hired a CEO and handed over management, I started Aigo.ai and have been focused on improving the technology and making it more relevant to commercialize the next generation of Aigo.  

 

What’s the role that AI-driven agents or bots can play for business?  

In the enterprise, the Chairman of SAP said that his software needed a co-pilot. Software is becoming more and more complex and it’s difficult for users to navigate. With a natural language interface, you can explore and navigate the software. You could tell the software to show sales in a given area, or run a cash flow report and run it again a week later with changes. There’s a lot of interest from robotics companies using natural language, from gaming or in the medical field for diabetes management. There is a wide spectrum in the enterprise.  

We are working to bring Aigo to the consumer to provide a personal personal assistant. With this model customers own the data and the models (this is different from Google, Amazon Alexa and Siri where corporations own the data). The second meaning is that the engine learns preferences and background information. The technology is much more advanced than what Siri and Alexa are using. 

Aigo Picture1

What have been some of the shortcomings in the field?

Gartner talks about the 3 waves of AI. The first is flow charts, the second wave hit about 5-6 years ago with big data and machine learning/deep learning. Originally people thought that if you captured enough data you could do natural language processing. For the third wave we need to move beyond static models into areas where there is cognition, which we’ve been working on. Big companies have been benefiting from big data statistical approaches in order to create profitable businesses so they have not invested in more advanced technologies. The cognitive architecture has not seen a lot of academic research or funding so far.

 

What is different about cognitive programming?

It’s a different way of thinking about the technology. Traditionally, the best programmers and engineers had an engineering mindset to solve the problem – they don’t think about the problem like a cognitive psychologist. Cognitive architecture deals with short-term and long term memory and reasoning.  

Current technology does not remember what you’ve said in the past. There are many challenges to deal with natural language. A 5-year-old child will learn concepts and build on them over time, none of the current systems are built like this. Demis Hassabis said “we need to look at our brain first”.  Geoff Hinton believes that we may need to throw away a lot of our old techniques and start anew. 

There are a number of technologies that work together in an integrated fashion in the new architecture, versus more of a linear process of today’s chatbots which have no memory, no ability to reason and remember.  This allows the next generation of bots to have better memory.

Picture2 Aigo

How does this differ from what we hear about machine learning?

The terminology of learning is confusing. When you apply algorithms to bulk data that’s one thing, but being able to learn interactively is a different thing.  Feeding a lot of data into an algorithm is very different from an interactive approach. At Aigo we are building an engine with core knowledge and skills, then there’s an application layer that can incorporate specific knowledge for a specific app. The outer ring is how an individual would do it. 

 

How can this next generation of cognitive bots change our relationship with computing?

Today, the prominent bots like Alexa and Siri do not remember the user’s preferences and history to be useful. There are too many bots and no one can remember what they all do – we think a lot of bots are going to go away. We think of a personal assistant as an exocortex for the brain which can talk to other bots. A true personal assistant provides the meta-layer that will orchestrate a request but interact with the owner in a way that’s highly personalized and stays with the user over time.

 

How are you employing blockchain in the technology?

Each Aigo is a unique instance with a serial number recorded on the blockchain. As people develop skills for Aigo they will be rewarded through smart contracts through the blockchain. The Aigo community will benefit from deflation. We have a utility token that people can use to purchase services and earn through contributions.

orange-line

 

Momenta Partners encompasses leading Strategic Advisory, Executive Search, and Investment practices. We’re the guiding hand behind leading industrials’ Digital Transformation strategies, over 100 IoT leadership placements, and 20+ young IoT disruptors. Schedule a free consultation to learn more about our Connected Industry practice.