Amelia is a new hire at a call center. She answers in two seconds, solves the problem in four minutes instead of an average of 15 minutes, and customers are quite happy. Amelia never goes home.
Amelia is what might be the most advance artificial intelligence so far on Earth. Developed by US based Ip Soft that normally sells autonomous systems for management of large it systems — you know, replacing human intervention.
For 15 years Ip Soft has been working on a secret side project, developing a cognitive system with the aim that it must:
1. understand natural language
2. learn through natural language
3. leverage what it has learnt to solve problems
The system was presented last month, and a few days ago Ip Soft visited Stockholm to show the technology to some 50 potential customers.
Proof of concept has already been performed with Amelia working in customer service or internal help desk at a handful of large American companies during the last year.
Gartner was allowed to talk to these companies and states that Amelia is the next level up from IBM Watson, the system that won over humans in Jeopardy in 2011 and now works, helping doctors to diagnose cancer. In the report Gartner says:
Gartner verified over 10 direct examples of IPsoft client references that all unanimously supported productivity benefits of a much higher magnitude (consistently over 50%) than any other managed services offerings in the industry. It’s rare to find a unanimous endorsement of this type. One client example had 56% of its client incidents resolved without human intervention and a 60% reduction in the mean time to resolution for its IT service desk.
One client — an oil and gas company — wanted to verify that Amelia could handle a helpdesk situation on an oil rig and let her digest a manual for a centrifugal pump for a few seconds. They then asked her questions.
“Which are the parts of the pump? What could a ticking sound be a sign of?”
I got a demonstration of the same test, and Amelia answered promptly without hesitating — short and concise answers.
Most impressive though is maybe how she learns. When she first started to work in call centers she often had to pass questions on to a human agent. But Amelia then stayed on the line, listening to the conversation, learning from it. And after 30 days she reached the efficiency levels Gartner reports.
“The best method for Amelia to learn is like for you, by doing things yourself. Amelia learns from interaction with people. During the first 30 days her learning is exponential as compared to the lab,” Ergun Ekici, co-founder and VP of emerging technologies at Ip Soft, told me.
The funny thing is that Ekici says he has a hard time believing that machines could one day become as intelligent as humans. He sure has a point though, saying that after 15 years of research in this field, he has grown a tremendous appreciation of human intelligence.
IP Soft says Amelia will be introduced commercially within a month or so.
CEO of the company is Chetan Dube. It’s privately financed by Dube’s family.
When will systems like this be used to analyze, develop or implement government policy?
@Fibb
hopefully soon 😉
The key take-away anyway is that this kind of technology evolves faster than many think.
Dear Mats,
In case you are collaborating with her, please tell Amelia that I want to hire her as a leader
of one army fighting against Probletence. See please
http://egooutpeters.blogspot.ro/2014/10/stop-probletence-pandemic.html
Obviously you are welcome too, please help disseminating this Manifesto – in Sweden too.
Stop Probletence! Stop Koalemos!
Your blogger colleague,
Peter
Impressive, scary and sobering development
Ask her where you can find an E Cat…or better when?