|
|
|
|
|
|
|
|
The 21st Century’s AI Arms Race Unfolds
Technology Briefing |
|
TranscriptFor several decades the fields of data science, machine learning, and artificial intelligence have been rising in prominence within the business community. The assumption has been that the better their ability to collect data, store data, and use that data to create actionable predictions of the future, the easier it will be for companies to survive, thrive, and prosper. Meanwhile, those without these intelligent capabilities will fail to flourish. That's why companies focused on leveraging AI to enhance business or health outcomes are sprouting up everywhere. Consider Livongo, an AI-based company in the healthcare space. It collects data, stores it, and creates predictions based on it with the intention of reducing costs and enabling better health outcomes for patients. From a technical standpoint, using data to predict health outcomes for humans via machine learning and AI is no different than using data to predict business outcomes via machine learning and AI. In the case of another company called Splunk, its Information platform allows data collection (through machine sensory input), storage (through databases & indexes), and predictions (through its machine learning toolkit); And this enables its customers to create businesses that are more capable of identifying inefficiencies, predicting future outcomes, and ultimately thriving and winning in their ever-evolving business eco-system. In short, Splunk makes businesses more capable of surviving. In response to this dynamic, we're beginning to see an arms race in the business community. That is, offerings from companies like Splunk and its competitors, including Elastic, are increasingly being seen as something like nuclear weapons for nation states. Why? Because without these weapons in their arsenals, companies will be rendered as impotent on the global business stage, as would-be superpowers without nuclear weapons, after the middle of the 20th century. In short, if they are to dominate their industries in a world where powerful machine learning and artificial intelligence are truly impacting business outcomes, businesses must acquire data analytics capabilities. A key driver of this new reality is the overwhelming growth of big data. As of 2019, over 2.5 quintillion bytes of data were created worldwide, every day. "There's no way around it: Big data just keeps getting bigger. The numbers are already staggering, but the growth is not slowing down. In 2020, it was estimated that for each person on earth, 1.7 MB of data were created every second." That's equivalent to every person on earth publishing a mp3 song every second of every day! As shown in the printable issue, in 2020 the cumulative quantity of stored data was as much a 100-times larger than it was in 1970 and it's growing at an accelerating rate. This avalanche consists of two kinds of data: structured and unstructured. Structured data includes the kinds of words and numbers you expect to find in a database. This kind of data is growing very rapidly because of the myriad sensors enabled by the Internet of Things. Unstructured data includes all sorts of video, audio, and print communications. Notably, the vast majority of the world's data is unstructured. And the mountain of unstructured data is growing at a truly exponentially rate, largely in the form of user-created content. As a result, companies are being bombarded by an explosion of data, and in order to handle it and actually leverage it as a company should, they must employ platforms, such as Splunk or Alteryx, or risk falling behind in the data analytics arms race mentioned earlier. For decades now, data analytics has been prophesized as potentially saving us from business inefficiencies and government waste. But thus far, there is just a just handful of public companies in the "pure data analytics" space. Meanwhile, even Amazon Web Services, Microsoft Azure, and Alphabet's GCP are collectively generating only about $100B in annual data analytics revenue, after 10 years. These market stats highlight just how immature the industry is and why it's likely to experience incredibly explosive growth, under the right circumstances. From its beginning, the data analytics gold rush was expected to occur in tandem with the maturing of artificial intelligence capabilities that would transform the way humans live and do business. Since Alan Turing, John von Neuman, and Norbert Weiner created the first inklings of computer science and artificial intelligence humans have awaited when AI would take the reins of business and begin generating so much wealth that we'd all be forced to drink from the golden spigot of the "government dole." Notably, Ray Kurzweil's best-seller "The Singularity is Near" pushed this thesis. However, history and science do not seem to support the idea that data analytics and AI will supplant most human workers in the 21st century. Nevertheless, the rise of these technologies will impact nearly every company in nearly every industry. The explosive growth in the market value of the FAANG stocks is the clearest example we have of what AI and big data can do to impact the "value proposition." Consider Facebook. Its access to billions of data sources has positioned it to be an early-stage "intelligent platform." The data fuels the intelligence which identifies and recommends, for instance, old friends from high school as people with whom you might want to connect. In the case of Google, the tens of exabytes of data the company has accumulated enables it to generate better and more fruitful search results. Because it essentially has "memories" of our past interactions with it, it can generate search results that are more tailored to what an individual, needs. It's like our interactions with other humans: Upon first meeting someone, we could never forecast what they might want to eat for breakfast. But after 50 meetings with that person, we could likely "live their lives for them." It's the same with big data, machine learning, and artificial intelligence: the key is accumulated and interpreted data. Prior to the 21st Century, data was sparse, but as the graph in the printable issue demonstrates, data has proliferated and continues to expand exponentially. The simple routine of data collection followed by iterative learning is the foundation of all that we do. Beginning in 2013, the year when Trend editors published RIDE THE WAVE and the current secular bull market in stocks began, "online searches for the terms "Data Science," "Data Analytics," and "Machine Learning" all exploded. These trends in search activity coincide very closely with the massive advancements that have been made in artificial intelligence. For over a decade now, companies, such as Splunk, Alteryx, and more recently Elastic have produced data analytic offerings. However, despite very compelling "use cases" and the utter explosion of data collection, these companies have not experienced geometric growth like Alphabet, Amazon, or Facebook. Why? Aggregation of data onto enormous cloud platforms was not widespread until barely 10 years ago. Now, through offering, such as AWS and cloud-based/ hybrid platforms, such as Splunk and Elastic, there is a centralization of data from, potentially, trillions of collection points all over the earth. Big hurdles including primitive data analytics, AI, and Machine Learning as well as problems with siloed data were only recently overcome. Before that companies couldn't identify patterns and generate forecasts from their data that were sufficiently valuable to warrant the required investment. But now, even though machine learning and AI are still in their infancy, we get closer every day to software and hardware capable of delivering cost-effective game changing solutions to crucial problems. But, even today, the many AI applications which process information for us still address so-called "narrow AI" use cases. Yet while so-called General AI has yet to be created, recent trends suggest that, machine learning capabilities are evolving to enable narrow AI to actually learn in a meaningful way. Consider "Google Translate." When introduced, It was completely useless if you really wanted to communicate with someone who spoke a different language. Today, it functions extraordinarily well, because it has had years to iteratively learn. This is just one of the simplest, many easily accessible demonstration of machine learning in action. It demonstrates that we can create software that actually learns, which means that the more it does or analyzes, the smarter and more accurate it becomes. So far, few companies have meaningfully leveraged data analytics, AI, and machine learning because its predictive ability has been too weak for many use cases. But with machine learning rapidly advancing to the point at which it's truly useful and big data access accelerates, companies will now be forced to adopt data analytics platforms so as to leverage this incredible technology. Consequently, business competition in the 2020s will become "a data analytics and machine learning arms race," just as business competition became "a factory optimization arms race" in the first-third of the 20th century. Given this trend, we offer the following forecasts for your consideration. First, by 2025 there will be a substantial performance bifurcation between the businesses that can implement the capabilities needed for usefully sifting through quadrillion of bytes of data daily and those that cannot implement such capabilities. The adoption of cloud-deployed platforms that aggregate data and allow users to create powerful machine learning applications is just beginning. For more and more industries, these capabilities will be part of the basic cost of doing business. Firms which can't reasonably expect to do so should formulate an exit strategy, such as being acquired. Second, the demand for platforms that are capable of monitoring, analyzing, and making decisions based on inconceivably large data sets will rapidly increase. At this time the use cases for big data and corresponding Machine Learning and Artificial Intelligence applications cannot be reliably identified or quantified. Each year myriad use cases emerge as more and more devices become intelligent and AI becomes better. Third, the AI arms race will represent a huge opportunity for "pure-play data analytics companies" that act as "off the shelf" solutions for companies which want to quickly create powerful artificial intelligence to power their businesses. Two leading candidates are Splunk and Elastic. Both companies were founded on the premise of searching through vast quantities of data and using that search capacity to create actionable insights for users. And today, both are creating machine learning algorithms that will continue to evolve into something resembling “general artificial intelligence,” which we expect govern most aspects of the businesses in which the solutions are deployed. |
|
Comments
|
|
|
|
|
|
|
|
|