Steve Brown - The Innovation Ultimatum

Artificial General Intelligence (AGI), also known as strong AI, describes AIs with the equivalent intelligence of a human being in any area of human expertise that you might imagine. An AGI can perform any intellectual task that a human can. Researchers continue to chip away at the problem of creating an AGI but currently have no clear plan for how such a feat might be achieved. AGIs aren't created by bolting together enough ANIs. It doesn't work that way. Never say never, but AGI, if we achieve it, is likely several decades in the future.

Artificial Super Intelligence (ASI) is where things get really exciting, or really scary, depending on your viewpoint. ASI defines an intelligence that surpasses the intellectual ability, knowledge, creativity, wisdom, and social skills of the very best human brains in almost every field. An ASI could be just 1% smarter or a million times smarter than the smartest human. In theory, an ASI machine that we build could design future, more powerful machines that operate in a way we aren't able to comprehend. Once ASI is achieved, rapid, runaway advances could follow. If that idea makes you uncomfortable, you're not alone. Exemplars of the high-tech and scientific world—including Elon Musk, Bill Gates, and Stephen Hawking—have all spoken out on the dangers of artificial super intelligence. It's hard not to think of the movie Terminator when you think about the professed dangers of AI becoming self-aware and designing better versions of itself. How engineers develop narrow AIs today may inform the design of super AIs in the future. This is why research and engineering focused on transparent and unbiased AIs is so vital.

Artificial intelligence holds huge potential and will drive successive waves of innovation throughout every industrial sector. As your organization starts to build out a comprehensive, multiyear AI strategy, here are a few places to start your thinking.

Use AI's predictive capabilities to forecast demand, streamline operations, target marketing efforts, predict trends, and influence the design of future products. Energy companies use AI to forecast demand for electricity. Fashion designers use AI to suggest fall colors for a couple of years from now. Equipment makers use AI to predict when their machines will fail so they can schedule preventative maintenance. What could AI predict for your business?

Future super sensors may improve the safety of autonomous vehicles, become “the next CT-scan” in medicine, or give us all a sixth and seventh sense. Super sensors will smooth business operations, revolutionize human interfaces, and transform the products and services of the future.

Every company should lead a strategic discussion around the possibilities super sensors present. Determine how your company can use super sensors to get eyes on your business, gain insight on operations in real time, and use those insights to make high-quality, data-driven business decisions. Inspired by Google Soli technology, consider how you could build super sensors into your products to improve the human–machine interface. How might you see your customers in new ways with super sensors? What other insights might super sensors reveal about your business?

Every business should have a digital voice strategy. Voice platforms offer customers a new way to interact with your brand, will complement customer support capabilities, and can be helpful for employees who use their hands to perform their work or don't work in traditional office settings.

First, determine the goals for voice platforms within your organization. Find areas where information delivered via voice interfaces improves worker efficiency or boosts their collaboration. Ask what conversational capabilities your brand should create for existing customers, and how voice might be used to attract new customers. Consider the role of interactive, talking online ads, e.g., “Click to ask me questions about our latest model!” Explore how digital voice agents can reduce support costs and open new sales channels.

Customers always want more choice. When brands give us more choice, they honor our desire to be recognized as individuals. Consumers expect to interact with brands through many channels: face-to-face, phone, web, mobile app, social media, virtual reality, interactive augmented reality objects, chatbots, voice agents, and more. Each brand will offer as many of these choices as makes sense for them. A voice interface is a new front door to your business, and voice agents are ambassadors of your company's brand. You must secure that front door the same way you secure your website. And you need to teach your ambassador to always put your best foot forward.

Businesses will need to train their chatbots to interact with a “tone” that is consistent with their brand voice. Marketing departments will need to define appropriate design parameters: Should the voice agent strike a casual or a formal tone? Calm and efficient, or excited to serve? Serious, or with a sense of humor? Rated G for everyone, or more PG-13? Voice agents for Disney, Harley Davidson, and Moët & Chandon will sound very different from one another.

Decide where to develop your voice agents: in-house, using third-party frameworks, or on existing voice platforms. Google, Amazon, Apple, and other tech giants enable developers to create apps for their voice services. Google calls these apps “actions” while Amazon calls them “skills.” The make-versus-buy decision will vary by brand. Most will build a presence on the major tech platforms. For competitive or security reasons, some brands may elect to build a dedicated voice agent that is embedded inside their apps and websites. Banks may prefer their own voice agents for security reasons. Bank of America's Erica virtual assistant is a good example.

The future of voice may become more open. Today's main voice platforms, Google Assistant and Amazon Alexa, resemble AOL and Compuserve, the walled garden online services of old. The rise of the internet, with open standard web pages and browsers, broke down the walls of the online services and opened up the online world to anyone who wanted to participate. Efforts like the Open Voice Network seek to build standards for online voice akin to HTML for webpages and enable brands to build their own voice capabilities. Time will tell if this will fell the initial duopoly of the tech giants.

When AIs are trained with data that represents examples of human behavior, they learn unwanted human biases that are reflected in that data.

Amazon receives a small mountain of job applications each week. To triage resumes, Amazon built an experimental AI to surface candidates worthy of an interview. There was only one problem. According to a report by Reuters, it quickly became clear that Amazon's AI was biased against women. The problem with the AI had nothing to do with the way it was coded. The AI reflected historical bias in Amazon's hiring process. The unconscious bias of hiring managers had been codified in the recommendations of the AI, which learned to favor male candidates. Amazon canceled the project before it was ever deployed and reviewed its hiring practices. Sometimes AI holds a mirror up to our own humanity.

We should hold our AIs to even higher standards than we hold ourselves, and should strive to create AIs that don't reflect negative human bias. The state of New Jersey built an AI as part of their plan to do away with the bail bonds system. Cash bail has been a part of the U.S. court system for centuries but is criticized for penalizing poorer defendants. According to a 2013 study by the Drug Policy Alliance of New Jersey, 75% of the New Jersey jail population were people awaiting trial. The average wait time for trial was 314 days. Almost 40% of the people awaiting trial were there because they couldn't afford $2,500 or less in bail. Under New Jersey law, everyone must be given bail, no matter the crime, so rich people always avoided jail and bail bondsmen made good money. In January 2017, the state of New Jersey replaced their bail system with an AI that created a public safety assessment (PSA) for each defendant. The PSA was used as a guide by judges. The PSA predicts the chance defendants will commit crime while awaiting trial and whether they will appear for their court date. The new system reduced the number of people in New Jersey jails awaiting trial by 30%. The AI was trained with information about 1.5 million previous defendants in 300 jurisdictions. Race and gender information was deliberately removed from the training data. Developers also removed all information on the name, education level, employment, income, and home address of defendants. Any of these data can be a proxy for race and gender and thus might disadvantage some demographic groups. The way we train our AIs matters. We should expect our AIs to operate ethically, fairly, and without bias. Leaders must ensure that their teams work hard to strip bias from all the AIs they build.