Listen to this articleThe basics:
- NJBIZ panel explores why AI represents a fundamentally different technology shift
- Experts cited rapid adoption, legal risk, security and governance as key challenges
- Panelists stressed education, guardrails and explainability for responsible AI use
- Discussion covered impacts on business strategy, workforce, education and regulation
From instant adoption to legal risk, education and guardrails, a panel of experts convened by NJBIZ recently discussed why artificial intelligence represents a fundamentally different technology shift — and why businesses can no longer afford to sit on the sidelines.
The informative Feb. 28 discussion, moderated by NJBIZ Chief Editor Jeffrey Kanige, featured:
In the opening round, Kanige asked where businesses and users stand with AI, as the panelists laid out what they are seeing in their work.
“The theme that seems to be running through here — is that this is different. This is not a search engine. This is a quantitively different thing that’s happening now than what has happened in the past,” said Kanige. “Am I right about that? This is something that really does require some different thought process.”
“Yes – not only that. If you look at the other historical tech revolutions throughout history, whether it was the railroad, telegraph, personal computing, cloud computing, there was so much infrastructure that needed to be implemented before that really saw real-world effects,” said Nesler. “And even in some cases, you had to invent the products to leverage those technologies to reap those benefits.
“This is really the first technology that we turned it on, and it’s instantly available to people on their phones, computers, smart glasses; and is growing exponentially by the hour.”
‘Like it or not’
Nesler noted there are big challenges to overcome, such as security and Health Insurance Portability and Accountability Act (HIPAA) compliance.
“But it’s coming whether we like it or not,” said Nesler. “So, let’s deal with it.”
Duston picked up from there to discuss the legal, regulatory implications of the technology. Kanige asked how much education he has to do in his work to get people to understand that side of it.
“It’s interesting because a lot of the same laws are going to apply, just to the new conduct,” said Duston. He described it as old law applied to new technology. “One example I think of is employment decision-making. Let’s say you use an AI to sort through resumes. The law is not going to change there. If your AI agent decides to exclude all people over a certain age or exclude resumes of all people of a certain race, you’re still going to be liable for discrimination.
“And ‘My AI bot made me do it,’ is not going to be an excuse any more than, ‘I had an employee who went off the rails and did this.’ So, the education piece is more like: Remember, the same thing that a human can’t do also can’t be done by an AI,” he continued.
[AI is] coming whether we like it or not. So, let’s deal with it.
– Doug Nesler, director of AI, Vertilocity
“I think the important difference is, if you have a human employee make a mistake like that, you go ask them what they did and why they did it. In AI, you sometimes have absolutely no ability to question it or look behind what it was doing. The outcome was bad, but you can’t explain it at all. So, I think the focus is going to be, moving forward, trying to figure out the explainability, traceability – why were these decisions made – and that sort of thing.”
How students use AI
Gulya weighed in from the education perspective. Kanige asked if he spends a lot of time with students who don’t want anything to do with it, trying to show them use cases, or with the people who are comfortable with it and trying to pull them back.
“I think both,” said Gulya. “And one of the things that I try to do is give my students a great deal of agency in terms of being able to choose, along with me, how much AI is part of their educational experience. And I am sort of in the middle ground – because for each of my classes, I do have a certain number of things where I do really ask students to use it.
“If I say – well if you’re opposed to it, we’ll work through that and everything and I’ll try to accommodate as much as I can. But I also feel like to have nuanced conversations about what AI means for education and what it means for the workforce, I think we need to use it at least a little bit and dabble with it,” Gulya explained.
Replay: Artificial Intelligence Panel Discussion
Click through to register to watch the full panel discussion!
“I think there are a couple of big challenges here. That conversation connects to what people have been talking about. The first thing is – this technology is developing so quickly, and my students know this – that a lot of students, they know ChatGPT, they know general generative AI programs. Then others will know something about agents. So, I’ll say that I don’t think I spend more time with one group over the other.
“But I try to get all of us to work together in a communal space. Have these open conversations. And they actually pull each other toward the center.”
Must-haves: Guardrails, security
Abboud weighed in and noted that nearly 100% of his company’s clients use AI in some way, shape or form. But only a handful of them are thinking about how to build those guardrails.
“We have clients that develop software. They’ve got AI built into their coding props,” said Abboud. “You get folks that are coding in many different languages and different computer systems and access to different things. And they’ve got this AI chatbot that can literally read their code, edit their code and manipulate it.”
He stressed that guardrails and security are imperatives.
“Regardless of how you’re implementing AI, whether it’s loosely – to guide conversations, build notes, whatever – or if you’re using as ubiquitously as a couple of our clients are, literally ingrained into the terminals. You really have to think about the guardrails,” said Abboud. “So, the answer is yes. It’s everywhere. I’ve never seen a faster deployment of it.”
Over 90 minutes, the conversation snaked through several key issues pertaining to AI, such as more on the legalities and regulatory issues and challenges, adoption in businesses, governance, data privacy, impacts on education and employment, and much more.
Where to start?
Kanige also took questions from the audience.
“I want to start with this one because every time we do an AI panel, we get something along these same lines. Even though it’s being adopted and it’s more ubiquitous, we still get these kinds of questions. If this is new to you, where’s the best place to start?” Kanige asked. “And relatedly, are there classes available to learn about various options and AI tools to tailor it to your business model? So, the question is, OK, all of you are making this sound great. I got this business – I’m in charge. What do I need to do?”
“It depends on the size of the organization,” said Abboud. “There’s not a general answer here. It’s got to be shaped for what you do; what your goals are. So, what are your goals broadly? Is your goal to grow the business? Is your goal more profitability? Do they mean the same? So, you have to start with just understanding the business and going from there.
“So, it’s a very consultative approach. It really depends on what industry you’re in. What are the regulatory risks? Those are some of the questions that we start asking clients when they come to us and say, ‘Hey, we want to adopt whatever it is in AI. We want to understand what the risks are, first, before we do it.’ And then educate the client on the risk,” Abboud continued. “And again, to use this term that I’ve been using a lot here today – is the guardrails. What’s the acceptable loss that you’re going to take?
“What does it mean if this gets out there? So, start by asking smart folks, like everyone here on this panel, is where I would start. Ask your peers in the industry – and where have they gone. Learn from their mistakes, is where I would also start.”
What can AI do for you?
“When the firm decided to put me in charge of this stuff, what I actually did was I put sort of the most popular AI chatbots on my phone and started talking to them about my new role,” said Duston. “Hey, I do this kind of law. This is my kind of company. This is the size we are. What stuff can you do for me, Claude? What stuff can you do for me, ChatGPT?
“And just saw what it answered. It’s actually really good. Most of them are really good at that creative thought process, like just help triggering ideas. That’s a safe way to use it,” he continued. “You’re not really telling it anything confidential. I did have to cross-examine one of them because it told me it could do legal research. And I was like, ‘Oh, how does that work – do you have access to a database of cases?’
“And it was like, ‘No, I don’t have that – I’m sorry, I don’t actually do that.’ So, take that with a grain of salt. But once you get some ideas of a general nature like that, then I agree with Michael. The best thing to do would be to talk to other people in the industry. Or, to your point, Jeff, even very small companies have trusted professionals, whether it’s their attorney, maybe their accountant. Like, ‘Hey, what are other people doing that you’re talking to?’ would be a great starting point.”
‘Evolve or die’
As the discussion wound down, Kanige had the panelists offer their final thoughts.
“I’ll just close with, AI is part of the workforce now and you have to accept that,” Abboud said. “And to ignore it is a big mistake. To overuse it is also a big mistake. So, you have to find that balance that’s right for you. I’m not a big policy guy. I don’t really believe in it as much. But you have to start there. I’m more creative. I want to get things done. But you really need to say, ‘Hey these are the policies that we have. These are the risks.’
“And if your organization isn’t talking about it today, it’s evolve or die,” Abboud continued. “Your competition is going to either sell it for cheaper, do it faster or better. And you’re not going to be around. So, you really need to start thinking about it. You need to start thinking about the risks and then the reward on the flip side.”
“One thing that comes to my mind that we didn’t talk about is, I find you have to be very careful how you ask these things questions,” said Duston. “Maybe it’s just my mind and what I do for a living. Cross-examining them could be really funny. But it’s also important to keep in mind; it goes with what I said earlier about not treating it like a human being.
“It really, really badly wants to tell you what you want to hear. And it’s the most susceptible to leading questions that I’ve ever seen. You can say – ‘That thing you just said is wrong,’ like ‘Yup, I was wrong.’ And you’re like, ‘No, I’m just testing you.’ And then it’ll go, ‘Oh I was right the first time,’” Duston explained. “Just in line with not treating it like a human being. If you say I’m trying to accomplish XYZ thing, please help me justify it. It’ll tell you everything you want to hear without ever identifying the contrary position.
“So, as Doug pointed out earlier, you probably want to ask it the opposing question and see what it says as well. Just in general, be very careful with the way you’re asking it to do things.”
Open conversations
Gulya was up next.
“The last thing I want to focus on is just the importance of doing exactly what we’re doing right now: having these conversations, cutting across silos,” said Gulya. “One of the things that I think is true is that, especially with technology like this where it cuts across everything – education, industry, personal life, everything else – we need to have very open conversations about it.
“I come from higher ed. So, we love our silos. We love our departments. We love to just stay isolated through our own thing. But we can’t do that. We have to talk to industry. And we have to talk to people about all these different experiences with AI,” Gulya continued. “One of my experiences over the break with the holidays was I went several days without saying a single thing about AI.
“I talked to people who had never heard of ChatGPT. My wife has never used the program. She’s never used it or any other AI program. And I think that’s actually really, really powerful. Because, especially when we’re in the AI space and we kind of live and breathe there, we lose touch with just how different experiences are.
“But having those conversations, searching out those people across industry, education, everything else, I think that’s the key, more than anything else.”
Final notes
Nesler closed the discussion out.
“On Jason’s point, I got to thank you all. Because my wife’s so sick of me talking about AI that she likes me getting it out to you guys,” said Nesler. “But one of the questions I think we kind of glossed over is where exec boards and everything should be concentrating this year. If you haven’t done it already, just reiterating what everyone else said – prompt engineering, thinking about tone, role, context, and all the just different techniques around that, clarity, eliminating pronouns.
“Eliminating contradictions in your prompt is a huge skill that everyone in your organization has to master. And then once you get there, organizing your data sets, optimizing for the AI engines. And then just everyone thinking that this is truly an engineering task at this point — that we know it’s going to make mistakes, and there’s things that it can’t do,” said Nesler.
“But if you go down those workflows and those processes, you’ll get to that spot that we can automate this. And once that’s done, we can build onto it and automate more and more and more. It will be an exciting time.”
