It was a busy week for Artificial Intelligence.
The headliner was the AI Summit at New York City’s Javits Center— offering something for everyone at a packed event, Dec. 5-6. Meanwhile, uptown at the 92nd Street Y, former Ambassador Wendy Sherman — while discussing the future of diplomacy with Madeline Albright and the Financial Times’s Gillian Tett — sounded the alarm… describing AI as a threat to world politics and global stability.
Then Google broke up the AI-isn’t-creative party by announcing that AlphaZero had mastered the intricacies of chess — from scratch — in about as long as it takes to drive from New York City to Washington, DC. It then proceeded to trounce all competition.
This all came into sharp focus at the Summit, with more than 3,000 attendees, 300 speakers, and 150 sessions cutting across an arc of case studies, academia, business insights, and technical solutions from virtually every industry sector.
The Summit is part cheerleading event, part informational, and part salesmanship… with a dash of apology. Indeed, AI’s shortcomings have gotten their fair share of headlines in 2018.
How much intelligence did it take — artificial or otherwise — to catch the numerous well-publicized gaffes, including Facebook’s inability to stop the spread of “fake news” on its platform? Or Uber’s self-driving SUV that struck and killed a pedestrian this spring in Arizona? Or the epic #chatbotfails that inevitably get posted on Reddit or Twitter — comedy gold for everyone except the companies whose reputations took a beating?
This was also the year where Christie’s auction house sold a pixellated AI-generated “portrait” for $430,000, raising more than a few eyebrows across the art world.
But slicing through the fog of novelty and hype, the Summit offered a serious perspective on the power of AI today, as well as what the future may have in store.
The fundamental purpose of Artificial Intelligence, much like my field of design, is to solve human problems. The most powerful demonstrations showed how AI can cull through this messy world of ours and create order out of chaos — curate and organize, classify and score — and surface patterns that enable flesh-and-blood people to make better decisions.
So what’s the state of Artificial Intelligence in 2018?
Financial institutions are already being disrupted, but not the way you think:
- There’s been a lot of progress over the past three years in Fintech. Alex Lyashok(WorkFusion) sees a huge opportunity in automating backend banking processes — nabbing money launderers, and speeding up money transfers and onboarding new accounts: “It’s a $1 trillion opportunity”. But for all the talk of banking and finance at the summit — a topic that got its own day-long track — I didn’t see a single session about AI for investment and stock trading strategies. I’d love to hear an insider’s perspective on the role of AI in the recent market swings. Is there a different conference for that?
There’s a gold rush for data:
- Mountains of data are being collected, but AI craves even more. The answer: connected networks. When multiple sources are pooled, synergies begin to happen that can yield even greater insights for optimizing processes and understanding customers. Sylvain Filippi, managing director of Virgin Racing, spoke about the power of “collective intelligence” that is enabling, for example, predictive parts failure in automobile racing as well as the aircraft industry.
The future of AI is winner-take-all:
- Ben Goertzel, showing off Hanson Robotics spirited creation @RealSophiaRobot, sees the dark side of AI becoming concentrated in fewer and fewer hands. He notes that AI-specific networks reinforce each other — leading to a winner-takes-all future where the rewards of AI go to the handful of corporations that are big enough and rich enough to afford large-scale data collection and analysis.
AI as a Service is already here:
- Amazon Web Services (presented by Dan Mbanga) is democratizing the playing field. AWS announced all kinds of enhancements to its platform including forecasting, recommendations, personalization, and understanding documents. This feels revolutionary to me — putting the power of AI in the hands of developers everywhere, who never before had access to such capabilities.
AI excels at organizing data, leading to better strategic decisions by humans:
- The legal field is ripe for automating. Kausheek Nandy (from Boehringer Ingelheim) discussed AI’s progress in automating the grunt work of legal discovery. Over in healthcare, AI is showing promise for enhancing medical images to help doctors and technicians to make faster and better diagnoses. AI is also moving into patient care, drug development, and into the field of advanced diagnostics. In doctor’s offices and at the bedside, Charles Elkan (Goldman Sachs) sees AI boosting the productivity of physicians.
- This highlights the complementary nature of the AI-human relationship. AI excels at shallow understanding. AI can build a platform of information that enables physicians to make decisions based on long chains of logic. In other words, humans still own deepunderstanding. Jon Oringa showed how the stock photography site Shutterstock uses machine learning to analyze not only an image’s content, but also “understand” an image’s context and composition. Compositional awareness lets users search for images that match specific layout criteria — such as a blank area where the headline could go — making designers across the country jump up for joy.
Mistrust is one of the biggies holding AI back from taking off in business:
- People just don’t trust AI to make good business decisions, says DryICE’s Clayton Ching.
AI is creating more good jobs:
- Workfusion’s Alex Lyashok predicts a convergence of software engineers and knowledge workers into a huge workforce, 1 billion strong, orchestrating “intelligent automation” in the new field of digital operations. While we wait for AI to achieve true human-like intelligence (and we may have to wait for a while longer), there’s still tremendous job growth potential in domain-specific AI applications. But jobs that involve low-skill and repetitive work are at risk — such as tagging keywords, in the case of digital images, or assessing the relevance of source material in legal research and discovery.
Thus, skills shortages are holding AI back:
- AI is creating so many jobs that recruitment has become a major headache — there’s a severe shortage of engineering skills and strategic expertise, says Clayton Ching. In the current market, attracting and retaining talent is difficult and costly. I can understand why tech companies so frequently raid each other’s teams for talent.
Artificial Intelligence is a force multiplier:
- Sowmya Gottipatti (NBCUniversal) talked about the twin challenges of managing the immediacy of news while delivering a myriad of video formats for the various digital platforms. “You can’t have breaking news take an hour to reach the digital platforms where people need it,” she says. But with AI, you can do it. AI can customize the content and also push it out — without additional manpower.
Sustainability… AI scales beautifully whereas human power does not:
- Fred Gerantabee shared how Coty uses AI to build comprehensive highly personalized user journeys — from digital to in-store — which require a level of maintenance that simply isn’t possible with manpower alone. In the customer service field, HelpShift is making waves using AI and bots to create “magical customer experiences”, says Abinash Tripathy. Bots flip the funnel, moving the customer service problem to the top by eliminating variance through automated chats and decision trees, and saving human interaction for the harder problems. Lucy Wang at Buzzfeed is applying AI to the challenge of automated publishing — in a helper role, by sorting appropriate content for each channel, identifying evergreen stories, and scheduling, among other things.
On the Future Labs stage, the conversation focused on four realms of the future: Vision, Voice Robotics and Language. These talks took a deep dive into conceptual (rather than applied) Artificial Intelligence. There’s a mind-boggling amount of R&D happening today, and powerful ideas are in the pipeline on both the hardware and software sides — including machine learning, natural language processing, and robotics. As these technologies mature, we’ll be seeing an abundance of applications and use cases feeding future AI Summits for years to come.
Despite myriad advances, AI is still in its infancy. And it shows. The topic has reached critical mass enough to support a Summit — but this is only its third year. During a break between sessions, I chatted with a couple of my AT&T colleagues about how it feels like this industry is still on the left side of the maturity model. The scene reminded us of the early days of Digital Asset Management. We’ve observed the DAM industry mature, and have seen conferences find their audience as vendors and business customers together coalesce around emerging themes. AI is already making a strong showing.