Fully automated support assistants or cognitive agents are coming soon and deserves to be discussed in their own terms. Their impact will be unique as compared to most other technologies. Cognitive agents need passionate champions who can highlight the positive impact they will bring, while also steering their continued development.
Amelia is a cognitive agent, built by the US company IPSoft. Used for the purpose of automated service support in a business type environment. She was built using a combination of Machine Learning, NLP/NLU (Natural language Processing, Natural language Understanding) and Text-To-Speach processing. At the moment she replaces humans for simple task completion and answers typical support questions that might be expected to occur in a call-center.
Amelia is compared by IPsoft’s charismatic founder and CEO Chetan Dube, to UBER in creating the “UBERIZATION” of service support. While UBER has been impactful in the market, it has shortcomings and not really comparable.
UBER has disrupted the traditional Taxi market, but not fundamentally changed how a journey is made. The bigger shift will come with the widespread use of self-driving cars. ‘Intelligence’ around a domain will be the differential element.
There is a growing discontent from UBER drivers, especially as they consider their increasingly disposable nature. Again, self-driving cars will accelerate this change. Taking out the human from the process changes the dynamic.
UBER might have a limited shelf-life. Once a journey in a self-driving car become as safe but cheaper than a human-driven car then public adoption will most likely be rapid. Competition from better technology will disrupt the UBERIZATION model.
UBER relies on a simple technology stack for task completion. IPSoft should over time create something that is more sophisticated, adaptable to new domains, with an element of ‘intelligence’.
IPsoft’s Amelia is potentially so revolutionary that she can and should stand by herself. Amelia should be seen as the disruptor to traditional service support, just as UBER was a disruptor to the established taxi service.
Dube gave a practical example to demonstrate how Amelia could be used in a common setting, that of finding a printer in an unfamiliar business location, using text commands only. While the example works it could be much stronger.
The end result to the user will not have a meaningful impact on their career, job or working situation.
The same task can easily be completed with a quick human-to-human conversation.
It is a one-off task with a limited energy saving possibility.
With a full network of data, I would want to see information that can really impact an individual’s situation. Proactive, not reactive should be the aim.
“Hi, this is Amelia and these are 3 things that will improve your day.”
This being said, it is important to realize the complexity of machine understanding in a task completion situation. Humans are by their very nature, hard to replace or augment. It is clear where this technology is heading even in the early days of development.
The future will be an exciting human-machine hybrid, taking the best qualities of each. We should make sure to ask better questions, ask what is possible and build smarter systems, never forgetting what is the ultimate aim of the technology. Amelia and other cognitive agents like her should be championed for their unique strengths and afforded the space to grow on their own terms.