Will General AI Ever Enter the Factory?

Whether it's reducing labor costs, shortening downtime, or increasing all-around productivity, a lot of discussion is happening around the potential of artificial intelligence (AI) to transform manufacturing.

But the rise of AI has also come with its own concerns—many around the future of jobs. On one extreme, AI is merely going to transform the way we work and usher in the next industrial revolution. On the opposite, we're all going to be panhandling once robots and software take over our jobs.

The reality is somewhere in between. Ahead of his talk at the Atlantic Design & Manufacturing Show in New York City, “How Will AI Change Manufacturing Over the Next 5 Years?,” Kayed Almasarweh, Watson and Cognitive IoT Solutions Lead at IBM, took the time to speak with Design News. Our discussion covered the current state of AI in manufacturing, the need for regulation, and the role we can expect AI to play in the factory of the future.

Design News: When we talk about AI in manufacturing, the subject always falls to jobs. Do you see a negative or positive impact for AI on manufacturing jobs? Why?

Kayed Almasarweh: It is hard to characterize it as negative or positive, but AI will definitely have major influence on the job market in all industries and in different ways. Some jobs have the potential of being eliminated; others will change and adapt in such ways that workers will have to perform them differently. And the pay from some jobs may go through some adjustments, due to the machines being able to do more of the intelligent tasks.

That said, there will be newer jobs created as well in areas such as creating new ways to deploy AI, training machines, and doing smart maintenance on them as well. All of this will depend on how fast we will get to machines that can learn by themselves in an “unsupervised fashion,” which is considered by many experts as the highest level of AI and the one with the most impact on all of us and the way we live our lives.
 
DN: Do recent developments like Google Duplex concern you? Are we overreaching when it comes to applications for AI?

KA: The application of Duplex is targeting certain use cases focused on natural language and trying to model human behaviors to the maximum possible. Speech recognition and conversations are key to the success of AI, so none of that is of much concern to me.

Are there risks for system abuse or failures that may produce undesirable outcomes? Of course there are. And that is why it is always better to beta test such systems in multiple types of environments and industries to take a reasonable amount of time to make sure that known risks are eliminated or minimized. I also think that a certain level of regulation, especially around human safety and information security, needs to evolve for some AI applications.

DN: Similarly, news recently broke that Amazon was licensing a facial recognition AI to law enforcement. It brings up an interesting issue regarding regulation. Do you think that we will need regulations around what tasks AI are allowed to perform in the manufacturing setting?

KA: I think we need regulations to a certain extent, meaning that we need to allow the market to play out and determine the usefulness of AI and its applications. But certain outcomes of deploying AI that may affect human safety and personal data security and privacy need to be closely watched and regulated. I hear and read some radical views on how dangerous AI may become, but I don’t subscribe to that as long as we, as a society, regulate around safety and security.
 
DN: We often talk about AI performing specific, niche tasks, particularly in manufacturing. Do you see a need for general artificial intelligence as well? Could this perhaps serve some function in manufacturing down the road?

KA: I engage with clients who manufacture things all the time and yes, the focus continues to be on using “Supervised Training” AI models and applying them to a set of well known manufacturing challenges: areas like predictive maintenance; predictive quality; production optimization to improve processes and overall manufacturing OEE; and visual and acoustic/audio insights to improve the overall quality and reduce unplanned downtime across the whole manufacturing operations.

I'm starting to see some clients, who have conquered the above mentioned challenges, now getting more creative and thinking outside the box and starting to design analytics algorithms based on machine and deep learning techniques to extend the base systems and do more with the data that they aggregated. That is where I see the next opportunity for general AI in manufacturing, where we can dynamically connect quality, maintenance, automation, and asset management systems together to squeeze more value out of AI.

RELATED ARTICLES:

DN: IBM has been a big name in AI for a while now, thanks to Watson. As more manufacturers embrace AI, do you think we will be seeing increasing competition in the AI space with companies offering AI for manufacturing? What do AI companies need to do to stay ahead of the curve in terms of providing services for manufacturing?

KA: Absolutely. IBM is investing big in the manufacturing space and, as you know, we are not new to this, but recognized leaders in cognitive and AI for years past. IoT enabled this acceleration of AI in manufacturing and that is where our suite of solutions called Watson IoT Industry Solutions provides a robust platform.

Our approach is to continue advancing AI to augment the human experience and democratize knowledge in ways that will make AI and cognitive technologies of value to everyone. IBM's strategy is to support open industry standards and embrace open source and open technologies whenever possible. That said, there will always be niche players who will do one thing so well—like maintenance, quality, or automation—and they have a place in the market. Our goal is to encourage competition that will benefit our clients while we continue to lead with innovative technologies and services to provide clients with the most return on their investment.

Kayed Almasarweh will be delivering a talk, “How Will AI Change Manufacturing Over the Next 5 Years?,” at the Atlantic Design & Manufacturing Show in New York City on June 16. For more information visit: AdvancedManufacturingNewYork.com.

Chris Wiltz is a senior editor at Design News covering emerging technologies, including VR/AR, AI, and robotics.