WK1: Applied Artificial Intelligence

A Handbook for Business Leaders by Mariya Yao, Marlene Jia, and Adelyn Zhou

“For all of you who build technology to make tomorrow better than today.”

Everyone knows (and hates) that marketers use the latest buzzwords, regardless of actual utilization, as a unique selling proposition for their company: artificial intelligence is certainly no exception. The authors provide a chuckle-worthy meme: “AI is like teenage sex. Everyone talks about it. Nobody knows how to do it. But everyone thinks everyone else is doing it, so they claim to do it too.”

Although big data and machine learning systems are still in their infancy, the hype is real. Promises of cutting costs and generating revenue are ubiquitous when CTO’s approach the board with an intimidating budget for artificial intelligence initiatives.

This book details the challenges, and opportunities, facing businesses for the upcoming AI paradigm.

One of the tools provided by the authors in The Machine Intelligence Continuum is a list of machine capabilities that help to understand the varying degrees of the functionality of intelligent systems.

1. Systems that Act: rule-based automata. These are systems that function according to some pre-defined script, using manually programmed if-then rules.
2. Systems that Predict: These systems are capable of analyzing data and use it to make probabilistic predictions. Predictions are simply mapping known information to unknown information, does not necessarily have to be a future event. Systems that Predict are powered by statistical power, thus are only as powerful as the quality of data being used.
3. Systems that Learn: These systems make predictions just as statistical systems do, but with less hand-engineering. Powered by machine learning and deep learning, they can perform tasks without being explicitly programmed to do so. Learning systems first require relevant data that can make a prediction about the world. This prediction paired with higher-level judgment is used to execute an action. This action and its subsequent outcome are refined using feedback that improves the systems prediction ability.
4. Systems that Create: A defining characteristic between humans and machines for decades has been creativity, but recent breakthroughs in neural networks are revitalizing computational creativity. These systems are able to create original writing, artwork, music, and even their own AI systems.
5. Systems that Relate: Emotional intelligence is being baked into intelligent systems as a use of consumer products like Amazon Echo’s Alexa permeate our daily lives. Sentiment analysis, also known as opinion mining or emotional AI, extracts and quantifies emotional states from our text, voice, facial expressions and body language. This allows the systems to respond empathetically and dynamically, just as our friends would.
6. Systems that Master: We humans are Systems that Master: an intelligent agent that is capable of constructing abstract concepts and strategic plans from sparse information. Systems that Master are an important part of Artificial General Intelligence, better known as the Singularity, as the ability to transfer data across domains is not yet a capability of our current AI systems.
7. Systems that Evolve: These systems exhibit superhuman intelligence and capabilities, such as the ability to change their own design and architecture based on changing conditions in their environments. Humans cannot simply augment our memory capacity by inserting more RAM or think faster by upgrading our processor. We are limited by our biological “wetware,” whereas Systems that Evolve can with ease. Self-evolving agents will be capable of ever-faster self-improving iterations that would lead to eventual superintelligence, better know as what keeps Elon up at night.

Another challenge posed by the authors is creating a benevolent AI: void of innate discrimination and bias that could propagate into larger and larger disparities of inequality. An emphasis on responsibility when creating AI systems are a common trope throughout the book, with hopes that the dangers are carefully considered moving forward.

Being subtitled, A Handbook for Business Leaders, this book details an enterprise AI strategy that touches on the importance of creating a culture ready for AI-adoption. As most businesses are not techno-centric, justifying a chunk of expenditure for a high-risk, high-reward investment is tantamount. The authors assert the challenge for businesses, an opportunity for students and workers, as there is a shortage of competent machine-learning engineers and data scientists. They claim there are about 100,000 capable of providing true value in AI and machine learning initiatives, most being poached by industry giants like Google, Facebook, Amazon, and Microsoft. There are plenty of online resources like Udemy and massive open online courses (MOOCs) that can help you seize this opportunity.

Machine Learning systems can improve every aspect of enterprise functions, a fact that has news media outlets comparing the AI revolution to that of a doomsday event. The reality is that, presently, machine learning systems are better than humans at rote tasks and streamlining processes; not yet capable of completely replacing human workers. They are able, however, to free up the cognitive load of your workers to commit themselves to more meaningful aspects of their job.

The applications for AI systems in enterprises are as follows (respective to specific domains):

1. General and Administrative: These roles are riddled with tedious, but critical tasks like manual data entry. As computers do not tire, excel at repetitive tasks and require no compensation, intelligent systems are prime candidates for domains like accounting, legal compliance, records maintenance, and general operations optimization.
2. Human Resources and Talent: Hiring good people that work well together are critical for the success of any company. AI systems can be better than humans at role-matching, streamlining the interview process, career planning with retention risk analysis, and other administrative functions. An exciting aspect of AI in HR is providing identity-blind resumé review and evaluation, producing a diverse and qualified candidate pool.
3. Business Intelligence and Analytics: Extrapolating meaning from data and making intelligent business decisions are a match made in heaven for machine learning systems. AI can help with wrangling useful, clean and relevant data for training intelligent systems, constructing efficient data structures, and providing valuable analytics to steer you and your company to market dominance.
4. Software Development: The next-generation of software development will be augmented by artificial intelligence for rapid prototyping, intelligent programming assistance, automatic analytics and error mitigation, code refactoring, precise project estimates and, of course, decision-making.
5. Marketing: This is a domain that has already shown promise using natural-language generation to optimize digital ads and put recommendation and personalization engines on steroids.
6. Sales: Machine learning systems augment sales tasks through optimizing and automating customer segmentation, lead qualification, sales development, and sales analytics to free up salespeople for more valuable and meaningful tasks like the pursuit of high-potential sales leads.
7. Customer Support: Automation of Customer Relationship Management (CRM) systems are a promising application of artificial intelligence, along with reducing customer churn and increasing customer lifetime value using socially intelligent systems.

A must-read playbook for business leaders who see the potential of machine learning and seek to leverage dynamic technology for increased productivity and quality of life.

Get the book // Happy Reading!