- Integrated Systems
- About Us
- Integrated Systems
- About Us
AI predictions for 2019: With great promise and growth also comes great responsibility
AI predictions for 2019 include strong growth, as well as the use of AI in groundbreaking projects, like ending world hunger by 2030. At the same time, there are some ethical issues to resolve.
One of the areas we think has the most potential for the future is AI, and we expect big developments there in the year ahead.
What are our AI predictions for 2019? To start, AI technologies will continue growing. According to Gartner, the business value flowing from this category has already raced 70 percent ahead of 2017 levels to $1.2 trillion in 2018. Much of the money organizations spend on AI goes to decision-support applications; in fact, Gartner estimated that solutions, like deep neural networks (DNNs), will comprise 36 percent of the market this year. Gartner further stated that AI-driven virtual agents, like chatbots, will "account for 46 percent of global AI-derived business value in 2018."
Looking to 2019, I suspect we will also see growth in segments that are relatively small today, like decision automation and smart products. As AI matures, 2019 will also be a year for critical, higher-level conversations about its potential use and application. AI is more than a technology: It's a powerful tool for business, society, and governments. It has the power to do great things, but it can also do harm if used improperly.
In short, 2019 will be a revealing year for AI.
Fighting world hunger with AI
One of our top AI predictions for 2019 is that the technology will help humanity tackle some of its most serious challenges. This may sound grandiose, but the potential is there, as long as we're willing to commit to it. HPE is stepping up to do just that: Our goal is to contribute to ending world hunger in the next 12 years by using the power of AI.
We recently announced our participation in the World Economic Forum's initiative to end world hunger by 2030 through the innovative application of technology. This took place at the forum's Sustainable Development Impact Summit. The forum had called on public and private organizations to join them in an open collaboration to find solutions that will eliminate food insecurity. The goal is to feed a growing population on a sustainable, inclusive basis.
It's certainly a compelling issue. According to the United Nations, about 815 million people worldwide are undernourished. By 2030, the world's population will grow to 8.6 billion. If nothing is done between now and then, that number of undernourished people will only grow.
Our belief is that technology and foresight will help the world solve this problem. There are no easy answers, of course, but at a fundamental level, the food problem is, in part, a data problem. For example, do we know the most suitable crops are grown in the right places? Perhaps the planet can support more food output if we knew what to plant where and when. Data about soil and weather, combined with agriculture genomics, may lead to greater crop yields. Food distribution systems may also have room for improvement, and AI should be able to offer better answers than the ones currently available.
As a first step, we are collaborating with Purdue University's College of Agriculture, which is a leader in enhancing food security and optimizing ag-tech around the world. With Purdue, we are working on disruptive technologies, like IoT sensors, that can enable digital precision in agriculture. Our president and CEO, Antonio Neri, is confident our partnership with the forum will "bring together the right minds, resources, and focus to achieve real-world change, now," reports AP News.
Confronting moral questions around AI
Just as AI has vast promise to improve the condition of humanity, the opposite is also true. In the wrong hands, or even with the best (but misguided) intentions, AI could become a negative force in society. Therefore, I predict the industry will increasingly focus on resolving the potential ethical dilemmas of AI in 2019.
HPE is already on the task, leading thought-provoking dialogues about the need for a moral code in the AI field. It's much needed, as examples of potential abuse and accidental misuse of AI abound. For instance, the justice system is starting to use AI-driven algorithms in sentencing. In this use case, a machine takes historical crime data to predict whether a criminal defendant will commit another crime in the future. The problem is the data in the system, as well as the assumptions built into the algorithm, might easily comprise a lot of bias from its creators.
Or, AI could find a cure for cancer, but it could also start picking who gets the cure and who doesn't. AI can solve crimes, but AI can also invade privacy and expose people to unlawful searches in ways we can only begin to imagine. It can unravel mysteries of the past, but AI can also reveal secrets no one wants publicized.
Given these risks, who decides what AI can and cannot do? This is the moment to discuss enforcing human checks and balances on machines that can affect people's lives. We need to educate a workforce whose jobs evolve with AI. It's time to start talking about making AI accessible to all socioeconomic classes. I suspect we'll start hearing a lot more about this in 2019.
Facing the human and compute limitations of AI
2019 will also further reveal a number of factors that could constrain AI's growth. People are one obstacle. Data scientists are hard to find, recruit, and retain. As Gartner put it, "Most organizations don't have enough data scientists consistently available throughout the business." The technology itself is adapting, however, making it easier for non-specialists to contribute to AI projects.
Compute capacity also places limitations on AI. AI requires extreme compute, I/O, and network functionality. It has to scale exponentially. Traditional IT infrastructure is not adequate. The SUSE Linux blog spoke to this difficulty, noting, "The more powerful AI grows, the more compute power it needs. Training for deep learning systems, for instance, involves mathematical vector operations that often result in the need for thousands of cores."
Moving ahead: HPC and AI convergence
High-performance computing (HPC) offers a viable solution to the compute deficit affecting AI. Going into 2019, HPE will continue to expand HPC offerings that can power AI. HPC delivers the performance sophisticated AI programming requires. We are pioneering new approaches to HPC and AI convergence—bringing together partnerships, practices, and our Apollo and SGI hardware to realize AI's potential.
No one can say exactly what's coming in the year ahead, but there will definitely be more AI developments to talk about in 2019. Stay tuned.
- Artificial intelligence: Adoption, new use cases will be the focus in 2019
- Want to know the future of technology? Sign up for weekly insights and resources
Meet Insight Experts blogger Bill Seidle, Group Manager of High Performance Computing & Artificial Intelligence Solutions. With over 22 years of marketing acumen in the highly competitive field of IT, Bill leads the HPE HPC & AI marketing team, focusing on delivering the best-in-class portfolio to help customers unleash their data insights faster.
Hewlett Packard Enterprise