Machine Learning
Research: J.C. Kuang
Machine Learning, a subset of artificial intelligence (AI), is a field of computer science dedicated to creating and programming computer systems that can approximate human capacity for reasoning and use knowledge collected from a variety of different situations and tasks to solve new problems.
Increases in computing power and storage space, as well as reduced hardware size, led to considerable growth for machine learning. Industry predictions suggest the current market (valued at $613 million in 2016) could rise to $3.7 billion in 2021. Larger firms, such as Google, Amazon, Facebook, Microsoft, and Intel, collectively acquired over 30 different startups in the AI/ machine learning sector in Q1 2017 alone.
Machine learning can rise to meet several challenges in news and information, entertainment, and positioning.
Entertainment
Machine learning creates considerable potential in gaming. Systems benefitting from several lifetimes worth of gameplay could develop uniquely “human” talents such as object permanence, pattern recognition, and adaptation to stimuli, becoming a near- indistinguishable simulation of a human opponent. Game developers utilizing machine learning could offer authentic multiplayer gameplay without relying on human players — a potentially lucrative opportunity considering the current and rising popularity of e-sports and massively-multiplayer games.
On a less glamorous note, machine learning could also bolster “smart” streaming platforms with better curation and recommendations for users. Already, sophisticated systems might analyze user-generated data to synthesize taste profiles and make predictions about consumption patterns. Future profiles may not be based only on viewing habits, but also incorporate geolocation, biometrics, or social media behavior to generate even more tailored experiences.
News and Information
Machine learning systems already write earnings reports or sports articles that increase the content output of organizations such as the AP while minimizing the potential for errors. But automated content is not limited to simple, formulaic stories. News organizations could also benefit from a computer’s ability to create other, more complex kinds of web friendly pieces, such as video or infographics. Deep neural networks, such as IBM’s Watson, have already proven capable of producing a cohesive video narrative in the form of a cinematic trailer. Moreover, a machine-learning system tailored for transcription and cataloguing could save a news organization considerable time and effort by creating more robust metadata for cataloguing and/or processing visual content.
AI also has the potential to handle curation efforts for organizations with a surplus of content, such as large legacy news outlets or social media platforms. Facebook, for instance, notably turned over control of its Trending Topics feed to a complex algorithm from a human editorial team. While this particular case was controversial, inevitable continued improvements to machine-learning systems may eventually make such decisions standard in the industry.
Positioning
With continued advancements in machine learning, digital personal assistants — Siri, Alexa, Google, Bixby — should become more adept at tailoring users’ feeds and overall experiences on digital platforms, both portable and home-based. In turn, users’ willingness to provide permissions and information to these corporations will increase, enabling brands to take full advantage of intelligently curated digital experiences while enjoying an increased presence and trust.
Machine learning is suited for synthesizing actionable insights from unfathomable quantities of data. These insights could generate hyper-personalized brand experiences tailored to perfection. Brands could use this data to target consumers with more specificity than ever before. For instance, an advertisement for
a household product might be primed by a user’s search history, and deployed upon their arrival at a retailer that carries said product. The tone of the ad’s content might be further influenced by a user’s mood, extrapolated from biometric data provided by a wearable device.
Compelling Early-stage Startups:
AdHawk, New York City
Analyzes social media and search data to generate recommendations that can be automatically implemented.
Factmata, London
Automates fact- checking of digital media content via machine learning and natural language processing.
SuperFan, Mumbai
Customized personalized chatbots for specific brands and clients to improve one- on-one customer engagement.
Additional Resources:
- Google’s Great AI Awakening / New York Times Magazine
- Machine Learning Driving Innovation at Amazon / Business Insider
- A.I. Revolutionaries: A Series / Red Hat
