Pittsburgh’s AI edge? Physical tools, leaders say — but people have to come first

For Pittsburgh, riding the next wave of AI isn’t about jumping on the ChatGPT hype cycle. Technologists want to lead the physical side of things, instead. 

While data centers and robots may be a big part of that, it’s human integration that’ll really make the innovation stand out, industry leaders said at Technical.ly’s RealLIST Connectors 2025 happy hour at Accenture’s Pittsburgh office in September. 

The region’s manufacturing strength and potential for government support, they said, can position Pittsburgh as a model for building AI that lives up to human expectations.

“The things that are missing that we’re not training people on are the social, psychological aspects of [AI], and the design aspects of it,” Carnegie Mellon University professor Majd Sakr, who also works as chief learning and research officer of Accenture LearnVantage, said at the September event. 

What Pittsburgh can get ahead on is building physical AI that understands and fits into the human context. That, he said, is the missing piece that could take the tech to the next level. 

“We have a lot of experience in building software and unfortunately, we need to develop a lot of capability in not only building hardware, but building hardware that has this intelligence capability to it, that can adapt to the environment that it’s in,” Sakr said. 

The hardware advantage

AI rebrands itself every few years, and right now it’s all about generative models, like ChatGPT or DeepSeek, according to Adela Wee, cofounder and chief innovation officer at electronics manufacturer Hellbender.

It’s a lucrative endeavor, but doesn’t account for Pennsylvania’s emphasis on providing the infrastructure to power the compute behind AI. And none of those software gains can come without the physical stuff, first. 

“Once you build [physical AI], then you can build all your software around it, and you can build all these ecosystems around it,” Wee said, “but it has to exist.”

Beyond just the physical infrastructure, Pittsburgh also has an established manufacturing scene that’ll help it get ahead and stand out from other tech that can be made wherever.

“Software is something that can be made in a garage,” said Benjamin Kostenbader, founder of design firm Dot Foundry. “Hardware is going to be the thing that Pittsburgh has another layer of access to.”

A little behind, but that’s a good thing

Pittsburgh’s other advantage is learning from the lessons of the ups and downs from other regions and companies to build better from the get-go.

Mark Conklin, president and COO at Free Market Health, has seen success with that firsthand. Working in a highly regulated field that forces the company to go slower, which means it knows what’s working well and what’s not before implementation, instead of trial and error.

“We have to respect the data that we manage as a platform on behalf of like the clients and stakeholders that we work with,” Conklin said. “So for us, it’s like, we can’t move fast and break things.”

It’s only useful if it doesn’t do more harm than good, and if people can use it effectively, naturally integrated into their lives. AI has to engage with a human in a way that feels integrated so that the two entities can work together, according to Sakr.

How do we get there? Removing the silos that build piece-by-piece apart from one another, to put the human use and effectiveness at the center, instead. 

“You need the math, you need the computer science, you need the robotics,” Sakr said, “but we are forgetting that for these things to be effective, they have to work in and they have to operate in the environment where human beings live and work.”