AI

'Will I look dumb?' Human-like virtual assistants can deter help-seeking

'Will I look dumb?' Human-like virtual assistants can deter help-seeking

[unable to retrieve full-text content]

[unable to retrieve full-text content]

Virtual assistants have become increasingly sophisticated — and more human-like — since the days when Clippy asked if you needed help with your document. These assistants are intended to make programs and apps easier to use, but research suggests that human-like virtual assistants may actually deter some people from seeking help on tasks that are supposed to measure achievement. 
Published at Thu, 04 Jan 2018 20:34:30 +0000

8 0

AI Trends

AI Trends

Within the  latest release of its Pega® Customer Decision Hub, Pegasystems has announced the ability for customers to deploy AI algorithms based on transparency thresholds. Called the T-Switch, the software is aimed at helping companies mitigate potential risks and maintain regulatory compliance while providing differentiated experiences to their customers.

Not all  AI models are built with the levels of transparency needed to fully explain how the AI made its decision. This is an issue in highly regulated industries. While some opaque AI algorithms may drive powerful performance, the complex logic behind these ‘black boxes’ needs the ability to be fully explained, especially when the model causes unintended actions.

Further raising the stakes, in May the General Data Protection Regulation (GDPR) mandates that businesses must be able to explain the logic behind AI models using European customer data to make decisions. Fines will be assessed at up to four percent of global revenues for non-compliance.

As part of the AI-powered Pega Customer Decision Hub, the T-Switch will allow organizations to set the appropriate thresholds for AI transparency or opaqueness. Businesses will be able to predefine these levels for each AI model using a simple sliding scale from one (most opaque) to five (most transparent). The transparency scores will help guide business users to build AI systems using the right models that meet their organization’s transparency requirements while still delivering exceptional customer experiences.

Business users will be able to control the transparency of their AI based on the models they choose to deploy to drive a desired outcome. For example, it is relatively low risk to use an opaque model for image recognition that helps flag older content assets still using outdated branding. Conversely, for financial institutions under strict regulations for the kinds of loans they offer customers, marketers may require highly transparent AI models to ensure they can demonstrate the resulting product offers are appropriate for the financial needs for each individual.

Pega Customer Decision Hub users will access the T-Switch and all other Pega AI tools in a newly centralized Business Control Center for AI. This allows users to create, import, adjust, and monitor their AI models to drive better business outcomes in a single dashboard. This includes models built using:

  • Pega Predictive Analytics Director, to rapidly develop models that accurately predict customer behaviors;
  • Pega Adaptive Decision Manager, to automatically adjust models on the fly based on customer actions;
  • Pega Text Analytics for text classification, sentiment analysis, and intent detection;
  • Any third-party AI models available via PMML

Pega Customer Decision Hub works in tandem with Pega’s suite of CRM applications for marketing, sales, and customer service to anticipate customer needs and provide highly contextual brand experiences for each individual customer.

The T-Swich capabilities will be available by the end of October with the release of the newest version of the  Pega Customer Decision Hub.

“With increasing amounts of regulations, nearly any global organization needs to ensure its AI systems provide the appropriate levels of transparency that allow businesses to explain how they use customer data,” said Dr. Rob Walker, vice president, decision management, Pegasystems, in a press release. “The T-Switch, built into the AI-powered Pega Customer Decision Hub, ensures organizations maintain compliance while leveraging the power of AI to provide exceptional experiences that increases customer satisfaction and the bottom line.”

Pegasystems is primarily focused on supporting marketing and salespeople as customers. “As we bring more and more of this data science and machine learning technology into how we interact with customers, there needs to be an understanding of the risks and ramifications of AI,” said Don Shuerman, CTO and VP of product marketing at Pegasystems, i an interview with AI Trends.  “The T or transparency switch, as organizations move beyond the hype of AI into actually engaging customers in the real world, will be an important part of the strategy.”

He added, “We want organizations to have clear business level control of their risk tolerance vs. value tolerance. The underlying engine can figure out the algorithms that stay inside the company’s risk tolerance.”

Companies might want to be able to in effect audit the sales trail. “We use AI to recommend actions to salespeople,: Shuerman said. “If you give the sales guy a way to see why you made that recommendation, they eventually will trust the engine. Transparency is important to gain the trust.”

Pegasystem sells either a cloud-based subscription service, or can install its software on-premise. “Many of our customers start doing things in the cloud to move fast, then many move on-premise because they want to have the “brain” closer to the core systems,” Shuerman said.

Looking forward, Shuerman said Pegasystems is researching systems that allow the customer to plug goals into the systems, which then generates the rules need to accomplish the goals. Also, research is going on into conversational chatbots, virtual assistants and natural language processing. This is outside the robotic automation side of the business, built on the 2016  acquisition of OpenSpan, a robotic process automation software supplier.

Learn more about Pegasystems.

Published at

15 0

Virtual twin in ten minutes

Virtual twin in ten minutes

[unable to retrieve full-text content]

[unable to retrieve full-text content]

Avatars — virtual persons: A new system makes it possible to practice and improve motion sequences by providing individualized feedback in real time. The system is embodied by a virtual person acting as a coach. In addition, users see themselves as avatars — virtual copies of themselves in the mirror of the virtual room. The creation of such personalized avatars used to take several days, but researchers have now developed an accelerated process.
Published at Tue, 02 Jan 2018 15:38:13 +0000

20 0

Researchers can now make neighborhood voting predictions from Google Street View images

Researchers can now make neighborhood voting predictions from Google Street View images

A Google street view car is displayed at the Google Inc. headquarters in Mountain View, California, U.S., on Wednesday, Oct. 13, 2010. Google, owner of the world’s most popular search engine, said third-quarter profit increased as businesses spent more on advertising for online consumers. Photographer: Tony Avelar/Bloomberg via Getty Images

In a sign that computers will be able to perform image analysis as fluently as text analysis, a group of Stanford-based researchers were able to make accurate predictions about neighborhood voting patterns based on millions of pictures collected from Google Street View, reports The New York Times. While other academic projects have used artificial intelligence to mine Google Street View for socioeconomic insights (such as Streetchange), this project is notable because of the vast quantity of images that its AI software processed.

Led by Stanford computer vision scientist Timnit Gebru, the team of researchers used software to analyze 50 million images of street scenes and location data. Their goal was to find data that could be used to predict demographic statistics at the zip code and precinct (which usually contain about 1,000 people) level.

From those images, they were able to glean information, including make and model, about 22 million cars, or 8% of all cars in the country, in 3,000 zip codes and 39,000 voting districts. After cross-referencing that data with information from other sources, including the Census Bureau’s American Community Survey and presidential election voting records, the researchers found that they were able to make accurate predictions about a neighborhood’s income, race, education and voting patterns.

In order to get their AI algorithms to classify cars accurately, the researchers trained it by recruiting hundreds of people from places like Mechanical Turk, as well as car experts, to identify vehicles in a sample of millions of pictures. In the end, their software was able to classify cars in 50 million images in just two weeks, a task the Times said would have taken a human expert 15 years to finish.

In an article published in the Proceedings of the National Academy of Sciences, the team wrote that their technology can supplement the American Community Survey, which costs more than $250 million each year to perform. Since the survey is also labor-intensive, with workers going door to door, that means smaller areas with populations of less than 65,000 are often overlooked. As technology improves, demographic statistics may eventually be updated in real time, though the researchers noted that policymakers will need to be careful to make sure data is collected only at the community level to safeguard individual privacy.

Read the source article at TechCrunch.

Published at Tue, 02 Jan 2018 15:03:00 +0000

19 0

Technique to allow AI to learn words in the flow of dialogue developed

Technique to allow AI to learn words in the flow of dialogue developed

[unable to retrieve full-text content]

[unable to retrieve full-text content]

A group of researchers has developed a new method for dialogue systems. This new method, lexical acquisition through implicit confirmation, is a method for a computer to acquire the category of an unknown word over multiple dialogues by confirming whether or not its predictions are correct in the flow of conversation.
Published at Tue, 26 Dec 2017 18:46:20 +0000

24 0

Quantum coupling

Quantum coupling

[unable to retrieve full-text content]

[unable to retrieve full-text content]

Physicists have linked atoms and superconductors in a key step towards new hardware for quantum computers and networks.
Published at Thu, 21 Dec 2017 17:27:20 +0000

18 0

Machine learning will change jobs

Machine learning will change jobs

[unable to retrieve full-text content]

[unable to retrieve full-text content]

Machine learning computer systems, which get better with experience, are poised to transform the economy much as steam engines and electricity have in the past. They can outperform people in a number of tasks, though they are unlikely to replace people in all jobs, suggest researchers.
Published at Thu, 21 Dec 2017 19:30:32 +0000

29 0

Memristors power quick-learning neural network

Memristors power quick-learning neural network

[unable to retrieve full-text content]

[unable to retrieve full-text content]

A new type of neural network made with memristors can dramatically improve the efficiency of teaching machines to think like humans. The network, called a reservoir computing system, could predict words before they are said during conversation, and help predict future outcomes based on the present.
Published at Fri, 22 Dec 2017 14:03:13 +0000

26 0