Creating the Future of AI Beyond Business
Dreamforce 2019, Salesforce.com’s user and developer conference, is held at the Moscone Convention Center and various hotels in San Francisco from November 18-22, 2019. (© Photo by Jakub Mosur Photography)
Earlier last week at Dreamforce 2019, Salesforce Chief Scientist, Richard Socher, gave a preview of the future of Artificial Intelligence (AI) and how its advances are changing the way businesses connect with their customers and making teams more productive.
Richard Socher, Chief Scientist at Salesforce Credit: Salesforce
The Dual State of AI
While we have yet to enter the golden age of AI, Socher drew parallels between AI and electricity, highlighting the many ways the technology is ready to be applied across a variety of business and societal needs. In the current “electricity state”, computer vision is now helping to automate stock management in the retail environment, while the beginnings of assisted driving technologies are helping to make roads progressively safer.
At the same time, we are still in the process of working towards what was referred to as the “Sci-fi state” of AI – a vision of a truly integrated AI, such as the ability to have a full, almost human-like conversation with an AI customer service representative or autonomous tractors that can automate the entire food production process.
This dual state of AI – the transition between the electricity and sci-fi state – is ready to have a huge impact on a variety of applications, and Socher believes that the best way to predict the future of AI is to create it. Live demonstrations of these advances were conducted by Socher’s team of scientists at the session, giving us a sneak peek on the future of AI.
The Camera as the New Keyboard
Bringing optical character recognition and natural language processing to CRM has been the focus of Victoria Lin, a Senior Research Scientist, who demonstrated how a small business owner, could bypass a tedious manual invoice entry and with a snap from her smartphone’s camera and convert an image of the document into a spreadsheet in seconds. Further amplifying this was the ability to democratise data access, through the simple use of natural language queries, turning any business into a data-driven organisation in minutes.
The Future is Conversational
With ever-growing pressures on labour, businesses’ customer engagement strategies will undoubtedly evolve in unique ways to offer rich, yet efficient interactions. Caiming Xiong, Senior Research Director, ordered a pizza on stage, showing how a conversational AI agent could be trained to authenticate, handle interruptions and process multiple tasks in a single conversation. Future “Sci-Fi” AI will use Natural Language Understanding models to advance today’s chatbots to detect customer’s feelings and intent, as well as extract key information, while working in tandem with Dialogue Management Engines, to determine and generate appropriate responses.
Can AI have Common Sense?
Research Scientist, Nazneen Rajani, summed up Salesforce’s AI research roadmap by emphasising that Explainability is the key to a trusted AI system. Enterprises must ensure that machines are trained to explain their predictions and reasoning. ‘Trust’, Salesforce believes, must the core focus of its AI research, and will be the key to helping people have more meaningful interactions, understand AI safely, take the appropriate actions needed. This trust in AI systems extends far beyond the business environment, due to the nature of AI as an omni-use technology and its potential as a force for good.
Counting Sharks with Einstein
Socher closed the session with an example of how computer vision is helping to keep California’s shores safe – protecting both humans and sharks through computer vision. By combining the use of drones, Salesforce’s field service lightning, and a dedicated computer vision model built on Einstein vision, he demonstrated how researchers were now able to identify and track the movement of Great White Sharks along the coast through a mobile application and respond as needed. In addition to being plugged into this newly available data stream, the project has brought together different sections of the community, from surf instructors, computer engineers, lifeguards and marine biologists, to have conversations about some of the most challenging issues facing the ocean today.