AI hits its inflection point
Subscribe to Business News.
ANALYSIS: The widespread uptake of artificial intelligence is going to be a game changer across a range of industries.
Better crop yields, retail data that predict demand and more accurate cancer diagnoses sound like totally different concepts, but they do share a common thread.
The link is artificial intelligence, and it’s set to change the world as profoundly as the telephone did more than 100 years ago.
Once the stuff of science fiction, and even the name of a sci-fi movie, artificial intelligence, or AI to use its abbreviation, is happening faster than most people realise and already goes far beyond driverless cars and driverless trains, which are perhaps the most common examples.
What started in research and development laboratories is moving into every facet of business, and a failure to understand AI carries a risk of suffering the same fate as businesses that have failed to embrace the internet.
Investment bankers are leading the transition of AI from research into business applications. They, better than most, know how to take an idea and turn it into money – in the case of AI by merging science with commerce to create entirely new businesses, or to provide a winning competitive edge.
In a 100-page business focused research paper, New York-based investment bank Goldman Sachs argues that previous waves of technology evolution described as AI were actually false starts. What’s happening now is something different, which Goldman Sachs describes as ‘an inflection point’, a game-changing turn.
“The reasons for the inflection range from the obvious (more and faster computers and an explosion of data) to the more nuanced, such as significant strides in deep learning and specialised hardware,” Goldman Sachs wrote in its analysis of AI.
“One of the more exciting aspects of the AI inflection is that real-world case studies abound.”
Examples cited by Goldman Sachs all have application in Western Australia, despite most of the research and product development occurring elsewhere.
Machine learning (devices that learn from examples), which offers the potential to increase crop yields and decrease fertiliser and irrigation costs, as well as assisting in the detection of crop and livestock diseases.
The data being gathered via internet trading is being leveraged to enhance inventory management, demand prediction, customer management and trend forecasting.
AI advantages are evolving in drug testing and development, hospital efficiency and image recognition that can improve the accuracy of cancer diagnosis.
AI is being applied to help lower costs and increase returns by analysing data faster to take advantage of profit opportunities.
“The broad application of AI leads us to the conclusion that it a needle-moving technology for the global economy and a driver behind improving productivity and ending the period of stagnant productivity growth in the US,” Goldman Sachs said
“We see AI affecting every corporation, industry, and segment of the economy in time.”
To back up that comment, and in keeping with its reputation as a moneymaking machine itself, Goldman Sachs takes its inquiry to a level beyond sci-fi and the gee-whiz factor by naming companies leading the rush into AI.
Top of the AI product development pecking order are a group that might be called the usual suspects, the companies already dominating global technology, including Google, Amazon, Apple, Microsoft, Facebook, IBM, Intel and Uber.
But Goldman Sachs has taken a step below the current leaders in technology and AI development to identify 150 private companies from around the world that are developing technologies to apply AI in real-world situations.
It is out of the smaller R&D and product development companies that the next generation of leaders should emerge, potentially with products to enhance your business – or put you out of business; they include the following.
• Affectiva, which is developing emotion recognition software and other facial expression measurements that fuel prediction analysis about human behaviour.
• Blue River Technology, which uses robotic systems that recognise plants and make decisions about which crop to plant and weeds to eliminate.
• Vicarious Systems, which is developing an AI algorithm to mimic the functions of the human brain.
• Zest Finance, which uses machine learning and data analytics to help banks make accurate credit provision decisions.
What large and small technology development companies are doing is applying the two basic rules of AI – machine learning and deep learning.
Machine learning is the application of algorithms (the mathematics used to process data), which means that machines learn from experience and examples rather than relying on predefined rules – such as telling the difference between and apple and and orange.
Deep learning is a branch of machine learning with layers of networks. Each layer solves a different aspect of a problem. Identifying a train would involve one layer seeing if the object has windows, with the next layer looking to see if it has wheels, with multiple layers of data leading to the recognition of a train – or in a medical example a certain type of cancer.
The more data processed through a high-speed computer, the more accurate the result. Tests by Harvard University and the radiology department of Massachusetts General Hospital show the improvement in recognition in medical imaging.
The number of times a CT (computed tomography) machine is shown a body part improves its accuracy, with that knowledge retained. A scan human brain was barely recognised (0.3 per cent accuracy) after five tries, but after that data was laid down over 200 scans, the accuracy rose to 98.44 per cent.
According to Goldman Sachs, the AI inflection point has not been reached because of a single technology or scientific discovery. It is more a case of a number of developments meshing into a new way of analysing, predicting and planning with the aid of the massive amounts of data being generated by computers, which enable machine learning from experience.
The three key changes that have brought AI to the attention of business as a tool for future growth are data, faster computing and better algorithms. The more data there are, the greater the accuracy of predicting an outcome.
Two graphs used by Goldman Sachs highlight the data and computing speed factors, with research from International Data Corporation and EMC (an arm of Dell Computers) forecasting a 141 per cent compound annual rate of growth in data generation over the five years from 2015 to 2020 when the annual output will be 44 zettabytes (see graph A).
The second and third graphs are the cost of speed and cost of computing with speed continuing to accelerate and cost continuing to fall.
Putting a dollar figure on the benefits of AI is not easy, but potential annual benefits by the year 2025 benefits (in the US alone), include:
• $US20 billion from optimising seed planting, fertiliser use, spraying and harvesting, sorting crops and identifying sick livestock;
• Between $US34 billion and $US43 billion in the finance sector by identifying better trading opportunities, cutting credit risk and monitoring email for compliance;
• $US54 billion by improving drug research and reducing trial failure and decreasing procedural costs;
• $US54 billion in cost savings for retail industry plus an expected $US41 billion in new annual revenue through enhancements such as predicting product demand, optimising pricing and enabling image-based product searching; and
• $US140 billion in cumulative cost savings for the energy sector by improving equipment reliability, streamlining project identification by fusing geological and production data and reduced maintenance downtime.
“AI is the apex technology of the information age,” Goldman Sachs said.
“The leap from computing built on the foundations of humans telling computers how to act, to computing built on the foundations of computers learnings how to act has significant implications for every industry.
“One of the more exciting aspects of the AI inflection is that real-world usage cases abound.
“Where large data sets are combined with powerful enough technology, value is being created and competitive advantage is being gained.”