Blog

The (Machine) Learning Curve: Understanding Artificial Intelligence and its Applications

The (Machine) Learning Curve: Understanding Artificial Intelligence and its Applications

 

“[Artificial Intelligence] is poised to transform business in ways we’ve not seen since the impact of computer technology in the late 20th century,”
– Accenture CTO, Paul Daugherty

 

Artificial Intelligence seeks to create systems that can sense, learn, think, and interact independently

Artificial intelligence (AI) is a buzz-worthy term that many throw out but few adequately define. From a conceptual lens, one might think of AI as the outer shell of a Russian nesting doll, a set of dolls in decreasing size placed one inside the other.

Artificial Intelligence Superset

AI, the first “doll,” serves as the overarching term that addresses all methods, algorithms, and technologies that allow computers to perform tasks which typically require human intelligence. Machine learning (ML) would serve as the next largest doll, a subset of AI with its own sub-applications. ML refers to the ability of machines to sift through volumes of data, adapt to new information, learn new concepts, and make predictions. It serves as the foundation for most AI talked about today. Terms such as computer vision and natural language processing are examples of ML applications.* Deep learning is an even smaller subset of AI and ML. Deep learning, the third doll, involves the use of multi-layered artificial neural networks to automatically learn representations from data. This technology is powerful, yet still in development. It will help power self-driving cars, intelligent voice assistants, and more.

Most current AI use cases are with narrow (or weak) AI, where the computers perform a very specific task. An example of narrow AI is Siri. iPhone users can ask Siri how to get to a restaurant or store, but Siri cannot answer when asked for help with decision making. Siri would need general AI to boast a full-range of question answering capabilities. However, general AI, or the mimicking of the spectrum of human cognitive functions, is found only in Hollywood movies and R&D labs at companies like Google and Microsoft. Additionally, most machine learning systems are currently “supervised systems,” meaning they require humans to act as guides to teach the computers which conclusions to make. Unsupervised systems would power ML without human input. Like general AI, this capability has not been realized. To summarize, computers do not have the capacity to replace humans, yet.

AI is necessary to make sense of and derive insights from the data generated in the digital age

The uptick in AI developments can be attributed to the vast amounts of data created in recent years, coupled with advancements in data processing. Insights from AI create value in four key areas:

  • Smarter R&D and forecasting: Industry data trains AI algorithms to make better forecasts, informing supply and demand
  • Improved production and maintenance: AI digests massive amounts of data from equipment, spots anomalies, diagnoses the problem, and predicts if a breakdown is likely
  • Targeted sales & marketing: ML helps retailers offer personalized recommendations based on previous purchases or activity
  • Greater customer experience: AI-powered facial or fingerprint recognition can be utilized in restaurants or stores when customers place an order

AI adoption is growing, but still in the early stages. Only about 20% of AI-aware companies currently use AI in a core business process. Most companies with some AI awareness are still in experimental or pilot phases. This will soon change, evidenced through reports that ~95% of the top 100 enterprise software companies anticipate incorporating one or more cognitive technologies by 2020.

As AI becomes integrated into business processes, clear economic leaders will emerge. PWC projects North America and China will see the most economic benefits from AI, with gains coming from labor productivity improvements and increased consumer demand. The U.S. is the early leader in AI. Most American AI-related deals occur in California, New York, and Massachusetts, respectively. However, the U.S.’s superior positioning will not be sustained. Only about 62% of deals in 2016 went to U.S. AI startups, down from about 80% in 2012. Chinese growth in this market is undeniable; from 11 disclosed deals in 2016 worth $263M to 41 disclosed deals in 2017 worth $1.64B. Moreover, about 50% of the $15.2B invested in AI startups in 2017 went to Chinese startups. China will soon surpass the U.S. as the global AI market leader.

Economic Impact of AI by 2035
Economic Impact of AI by 2035

AI stands to change the business processes and raise economic output for all industries. Accenture projects the Manufacturing, Professional Services, Retail, and Financial Services industries will experience the most growth by 2035. Each industry holds a highest potential use case, with the use cases for the select four industries highlighted below:

  • Information & Communication (3.4 – 4.8% growth): Media archiving and search
  • Financial Services (2.4 – 4.3% growth): Personalized financial planning
  • Manufacturing (2.1 – 4.4% growth): Increased monitoring and auto-correction of processes
  • Wholesale & Retail (2.0 – 4.0% growth): Personalized design and production
Industry Output from AI
Additional Industry Output from AI

Historic investments in AI most often came in the form of internal R&D for the tech giants. Yet, external funding for AI startups has grown tremendously in recent years. Corporate M&A is now the fastest growing external source of funding for AI startups, with an ~80% CAGR from 2013-2016 and $21.3B in AI-related M&A completed in 2017.

AI-focused M&A transactions
Number of AI-focused M&A transactions

There are four main drivers for AI M&A activity:

  1. Acquihire (buying a company for the skills and knowledge of the employees)
  2. Bolster existing services
  3. Gain AI capabilities for the first time
  4. Develop general AI

Investments from VCs, PE firms, and other forms of seed funding also rose. In 2017, investors poured over $15.2B in funding to AI startups, a 141% funding increase from 2016.

Capture 5
Number and amount of equity financings

While PE firms tend to invest in AI-related companies, VC firms have focused more on AI technology offerings. Machine learning startups received the most investment attention, with $6B in external investments in 2016.

External investment in AI
External investment in AI by category

Company profiles highlight investments in AI, applications of AI

Two areas with noteworthy AI applications are media intelligence and communication intelligence. Media intelligence involves the use of AI to derive insights from large amounts of data from various media channels. Communication intelligence involves improving the way information is delivered and providing better insights for decision making.

Meltwater is one company active in the media intelligence space. Meltwater’s core product offering is Outside Insight, a web-based platform that offers insights regarding media coverage to four customer segments: PR & Communications, Marketing, Executives, and Enterprise. Meltwater’s impressive customer base ranges from Tesla to Netflix, Shell, BlackRock, and Coca Cola. Since early 2017, the company has made 7 AI-related acquisitions in areas ranging from natural language processing to data extraction.

Notable companies within the communication intelligence space include Automated Insights and Element Data. Automated Insights offers Wordsmith, a natural language generation platform that converts raw data into narratives. Wherever there is data, Wordsmith can write about any topic using any tone. The platform produces over 3,000 articles each quarter, and has a client base that includes the Associated Press, Yahoo, and Nvidia.

Element Data offers the Decision Cloud, a cognitive computing platform that integrates human-like analysis of options and trade-offs. Decision Cloud identifies potential customers, enables customer engagement, and facilitates decision making. Element Data has made three AI-related acquisitions since 2017. While still in the early stages, Decision Cloud has the potential to help retailers connect customers with the right products based on their needs and interests. Also, it could be integrated into existing intelligent agents. Referring to one of the first examples in this blog, with this integration, Siri could able to help users with decision making.

AI transforms how humans interact with data

 

AI augments the human role in a business and enhances human-computer collaboration. The main applications of this augmentation are as follows:

  • Conversational Experiences: Individualized, virtual conversations with customers
  • Real-time Personalization: Delivery of tailored content and promotions
  • Identity Resolution: Confirmation that the right message reaches the right person
  • Marketing Orchestration: Automation of certain arduous channels to increase efficiency
  • Augmented Analytics: Wider application of easier-to-use analytical capabilities

M&A activity will continue to surge as more companies seek to integrate AI into their core business processes. As tech giants push towards achieving general AI, smaller players should find success in niche markets. Enterprise level AI adoption will become essential in maintaining competitive parity. Machine learning, once a trendy buzzword, will become the norm. Looking forward, deep learning is the next focus for developers and investors alike. In conclusion, Artificial Intelligence is a broad term with many subsets and expansive business applications. Most importantly, it’s only just in its early stages of impact.

 

 

*Additional vocabulary:

  • Natural Language Processing (NLP): A computer system’s ability to understand and interpret human language the way that it is written or spoken
  • Ex: Google search and (“Did you mean…” response)
  • Semantic Analysis: A part of NLP – how machines identify the basic, logical form meaning of sentences
  • Ex: If you search the term ‘jaguar,’ and you might get results for a luxury car, a large feline predator, or a football team. Semantic analysis uses contextual clues and other content to determine what is the best match for your search.
  • Natural Language Generation (NLG): The process of developing a machine capable of sorting through variables and putting them together into natural, human-sounding sentences or statements
  • NLP ‘reads’ while NLG ‘writes’
  • Computer Vision: The extraction, analysis, and understanding of useful information from a single image or sequence of images

 

Works Consulted

CB Insights:

Accenture:

PWC:

McKinsey:

Nvidia:

HBR:

Visual Capitalist:

Other Select Company Websites

Request a Demo