Making the Leap: 5 AI Takeaways from The Economist Events Innovation Summit

March 22, 2018 • Blog Post • Kirk Bresniker, Chief Architect, Hewlett Packard Labs & HPE Fellow


  • Participating in a breakfast panel at The Economist Events’ Innovation Summit 2018, HPE's Kirk Bresniker met with the world's leading minds in AI to discuss the promises and peril the technology could bring
  • One of the key highlights from the event was a focus on tuning out the hype around AI and instead focusing on building technology that could take AI to the next step

Hewlett Packard Labs Chief Architect Kirk Bresniker shares his thoughts on the opportunities, obstacles and ethics of AI

Today I participated in a breakfast panel at the Economist Events’ Innovation Summit where we discussed a future in which enterprises, and the world at large, are powered by artificial intelligence.


We are at the peak of thought experiments around how AI could become ingrained in nearly every aspect of our lives, from businesses to scientific research to our personal interactions. It will be embedded in everything from the cars we drive to the phones in our pockets. But the question on everyone’s mind today was: “What is the first step, and how do we take it?”


After some riveting and relatively optimistic discussion with my fellow panelists, I was struck by how, despite being in an industry where things seem to evolve almost hourly, so much still needs to change and to be considered before we can achieve our nearer term AI goals.

Here are just a few of my takeaways from the morning of great conversation and learning:


  1. We need a reality check.

We have ambitious goals for an AI-powered future. Those goals keep us driven and inspired, but we need to be realistic because the gap between the hype and the reality is probably a lot wider than you think. The grandiose vision of AI delivered globally, fairly and sustainably cannot be achieved with the technology we have available today, and it will take real dedication and resources to get us there.


  1. We need a culture shift.

Considering how often we hear about AI being used by businesses, you’d think it was being used by everyone, everywhere, all the time. In reality, the adoption rate amongst CIOs is only around 4 percent, according to a recent Gartner survey. To take the leap from talk to action, a few things need to happen, not the least of which is a commitment to collecting data because a system is only as good as the data you feed it. Data stewardship is a skill you need to develop by doing. This all starts with clear direction from leadership, but requires employees at all levels to embrace the technology. AI is going to be an incredible tool for disruption that will favor the upstart because of its ability to turn data into economic advantage.  CIOs, have you sponsored your AI hackathon yet?

  1. Using the right tool for the job is paramount.

AI is a tool. And, it’s only as effective as the training it receives and the data it’s fed. It’s not enough to create purpose-built systems for individual use cases.  We need to nurture them with the training and information needed to provide actionable insight instead of reinforcing our biases. Fortunately, open source tools are becoming more accessible making it easier for businesses to create the trusted, reliable training infrastructure that will create high-quality systems.


  1. Bias in means bias out.

Intelligence craves data. If you have control over the data, you have control over the intelligence, but it’s unfair to assume the system will come up with all the optimal solutions with that data alone. Every time you throw away data or even worse, when you give away data to someone else, you are giving up on the long term economic potential that could be realized by an AI system.  AI can find correlations that you may have never considered, but only if you present it with the unreduced, un-redacted, high-fidelity data.


  1. Energy consumption cannot be ignored.

When it comes to AI, we face the real possibility of being victims of success. The faster and more powerful our computational abilities, the more energy we’ll need to fuel it. AI could be incredibly effective on the benchtop, but as we scale applications and create more complex models or the continuous operation, it will begin to overtake our ability to power it. There simply won’t be enough energy.  It’s a fact that’s often conveniently left out of conversations around the progress and promise of AI. These applications take massive amounts of energy to manufacture and train. But because that promise is so great, we cannot slow down its development, but rather, we must find ways to operate more efficiently. Luckily, HPE is already on it — read CEO Antonio Neri's perspective here.


The final thought I had was from the first question that came up from the crowd, “When will this technology be ready for me to start?” The answer is that you can—and you should—start now.  People are experimenting with AI in every business sector, including yours.  The only question is, “is it happening in a startup? At your long-time competitor? In your own organization?”


Watch the live stream replay below for more insights from the panel discussion, and visit our website to learn more about HPE’s latest deep learning solutions.


For Efficiency’s Sake: It’s About More Than Speed, It’s About Sustainability


A Universe Full of Opportunities, for AI and Supercomputing

Blog Post

HPE Launches Vertical AI Solutions, Dramatically Accelerates Deep Learning Training

Press Release