Many of the highest profile companies with the best performing stocks over the past two years including NVIDIA, Amazon.com and Microsoft have been world leaders in the research and development of artificial intelligence (Al). From speech recognition software, like Amazon’s Alexa, to Intel’s summer 2017 $15 billion purchase of Mobileye, an Israeli manufacturer of vision sensors for autonomous vehicles, Al-related products and concepts have begun to infuse consumer and business environments.
Al is an area of computer science that focuses on the creation of machines that can think and learn like humans. These machines can understand and speak language, recognize patterns and solve problems. Al machines create phenomenal new applications but also can cause significant disruption as they evolve. Al is based on three main components: machine learning, deep learning and neural networks. Machine learning is usually based on “trial learning” wherein machines are given large data sets and asked to repeat a task in order to achieve an objective. For example, a machine may be fed millions of images to analyze. After going through endless repetitions, the machine acquires the ability to recognize a pattern, shape or human face.
Deep learning involves machines training or learning from their mistakes. The process involves feeding a computer program massive data sets and asking it to make decisions about other data. The learning can be supervised, semi-supervised or unsupervised. Deep learning models are based on information processing or biological functioning, like human brains. Generally, some form of a neural network is developed to conduct the learning and training. A neural network is organized as a layer made up of interconnected nodes or decision points. The system learns from the output of each layer. The “deep” refers to the number of layers through which the data is transformed. Similar to a young child learning to speak, the neural net is supposed to learn how to process information via practice, trial and error.
Consumers engage with Al every day at home, at work and at play including Apple’s Sid and Amazon’s Alexa. While vehicles may not be fully autonomous yet semi-autonomous driving fleets are on the roads. Manufacturers like Tesla have connected all their vehicles with knowledge learned by one car shared across all models. Shared experiences provide a critical feedback loop in the machine learning process.
Within social media, Al works through users’ past web searches and interactions, to deliver a customized experience. Music and media streaming services like Spotify, Netflix and YouTube deliver introductions or recommended lists of new or related content based on current or previous selections. Gamers, whether on PUBG or Fortnite often play against Al powered “bots” while in some games such as Middle Earth, the Al enemies evolve based on their interaction with users. Online advertising networks and navigation and travel applications such as Google Maps or Uber were early adopters of Al technologies. Now the banking and finance industries are heavy Al users in areas of customer service and fraud protection with AI-driven emails alerting users when an unusual transaction occurs.
Looking forward, Al will likely incentivize firms to rethink their enterprise structures for sales, cyber-security, procurement, logistics and human resources. While the advancement of Al seems certain, predictions relating to workers, jobs and training are far less certain. While AI-driven technology may displace or eliminate jobs, it should not eliminate work In the near term, machine learning is most adept at replacing individual tasks. The authors of a recent paper in the American Economic Association state “full automation will be less significant than the reengineering of processes and the reorganization of tasks” leaving workers with more time for higher level tasks. While workers will need to adapt and mid-career training will become more important, difficulty in retraining may grow as the skills gap continues to widen. For many workers, more specific and advanced education may be required as roles and responsibilities evolve alongside smart machines.
Machine learning and automation will probably have lesser impacts on jobs that involve managing people, creative and design functions, technology professionals and occupations involving human or social interactions such as veterinarian, lawyer, pharmacist, nurse, school teacher, surgeon childcare and eldercare. Jobs conducted in unstructured environments such as gardeners or plumbers may also be minimally impacted. Globally, Al and machine learning may have drastically different impacts. Advanced economies may be more affected than developing ones, driven in part by differing wage levels, demand growth, industry mixes and demographics.
Predicting the long-term scope of economic change is notoriously difficult, especially given today’s rapid pace of technological change, enormous sums of capital investing in research and the speed with which new products penetrate markets. Throughout history, automation has consistently driven productivity higher, in turn driving profits and incomes higher. Higher incomes lead to greater demand for goods and services which then create newer or different jobs. This cycle continues today, only at a much faster pace and simultaneously across multiple industries and global regions. While the potential for short-term disruption from Al may be greater, the probability of unheralded innovation and growth is staggeringly higher.
Read the full October Market Review.
RCB Bank Trust offers free portfolio reviews at no cost, no obligation. We’d be happy to take a look at your current portfolio and offer a second opinion to ensure you’re getting the most out of your investments.
Connect with an RCB Bank Trust Wealth Advisor in your area.