Wrapping Up Triangle Talks - Demystifying Generative AI

At the end of July, I had the pleasure of hosting a lively panel in front of an enthusiastic audience in our new Slalom Raleigh office. We wanted to give people in the Triangle community a realistic perspective on what’s going on with Generative AI. Our goal was to help them see past the hype to understand how they could sort out its potential impact on their organizations.

Generative AI is a complex topic filled with technical, social, and political nuance, so I went back to my roots as a change agent. When helping people understand something complex, it helps to use simple frameworks. As the moderator, I asked the panel to walk us through their personal versions of “why, where, what, and how” they’re innovating with Generative AI.

THE WHY:

I started with,?“Why are you jumping into Generative AI now? What makes this the right time?”?For me, the answer is rooted in a lifetime of seeing technology transitions. Playing with an Apple II in 2nd?grade was incredible. Making the LOGO turtle draw designs or writing simple games in BASIC felt like living in the future. When I was in college and downloaded an early copy of the Mosaic web browser, I could pull information and content from places around the world without being a network engineer. The world again felt different. Years later, the emergence of “web 2.0” and the social web made it so much easier to share content with friends and family without operating a webserver or knowing HTML. The world felt different. The web became more democratized and participatory.

That’s how I feel when I experiment with various Generative AI tools and services. Without having to be a data scientist with deep knowledge of the underlying neural networks and training data/techniques, I can produce remarkable results. These models, and the transformers enabling people to modify and extend them are making AI and ML more democratized and participatory. The world feels different again.

The panelists had additional motivations. Some were fascinated by the psychology of how quickly people grow to trust algorithms that produce results in very “friendly” terms. Of course, he pointed out, that’s exactly what a large language model (LLM) does. It predicts the most likely words to make the user happy – a virtual “trust engine.” Others jumped into the Generative AI space to disrupt the status quo before being disrupted by it. Better to experiment and get ahead of the wave versus being behind/under it. My colleague from the public sector spoke eloquently of the opportunity to improve the citizen experience with government while also greatly improving processes and quality-of-work for state workers.

THE WHERE & WHAT:

All the “why’s” then led to the “where/what” portion of the framework.?I asked the panelists to describe where in their organization they were applying this exciting technology, and what were they asking it to do.??I’ve seen three distinct legs of early exploration. The first is in?super-charging customer interactions. The LLM’s ability to produce orders-of-magnitude better interactions with customers than traditional intent-and-slot chatbots is remarkable. The second is in?content creation/summarization. The ability for Generative AI to either produce new content at our fingertips or summarize massive volumes for easy consumption is astonishing. The third is as a?bionic assistant for developers. Experiments are showing significant gains in coders’ productivity and satisfaction when working alongside a smart agent that can suggest improvements, identify potential errors/vulnerabilities, and even generate entire blocks of code alongside them.

The panelists pointed out the value of intelligent agents in the sales process.?Helping sellers with better research, more integrated communications, and deeper integration of tools and processes are clearly areas of exploration for a variety of companies.?Any process where agents can retrieve information from structured and unstructured data sources to synthesize that information to guide conversation can benefit from Generative AI. Another panelist was looking at the entirety of the RFP process in his organization. He is exploring how LLM’s help standardize templates/language in RFP’s along with summarizing responses and researching response details. My last panelist works to secure AI and ML services. His firm ensures the data and tools used to create these amazing services are properly secured. When he tipped the discussion to security, we quickly pivoted to “the how.”

THE HOW:

I asked panelists to give advice to someone just getting started experimenting with Generative AI.?If they had a time machine and could go back 3 or 6 months, what did they wish they would have known??The variety of responses from the four panelists was noteworthy. To a person, they recommended?getting some “seat time” with these technologies. A flood of freely available tools let you try things out. Have a chat with a language model or turn a selfie into a superhero avatar of yourself. The second bit of salient advice was to?focus on peoples’ experiences. Adoption of new technologies requires a uniquely human focus on how people build trust in systems and how those systems impact their daily lives. Another panelist encouraged getting ahead of policy and thinking about how to responsibly apply Generative AI in your organization. While the frameworks and proposals are evolving rapidly, my panelists warned against a “head in the sand” stance of ignoring it or banning it altogether. As technologists, we have an obligation to put the right policy guidelines and technical guardrails in place to allow people to enhance their productivity without exposing data/secrets or misusing AI output. Simplifying the nature and use of these tools will allow us to engage people, as will explaining the difference of private and public tools and systems. As one wise panelist pointed out,?our obligation to help users understand where/when to share information isn’t anything new. We had to evolve these policies when we got PC’s. We updated them again when we connected those PC’s to the Internet, and further refined them as social media emerged. We will have to take that same next step with this new technology. It is on all of us to form policy and security guidelines and guardrails that allow for experimentation without introducing undue risk.

That's why I'm so excited to be at Slalom.?We have a uniquely (and fiercely) human perspective on these technologies.We've been practicing responsible AI for years, and we're in a fantastic position to help guide our clients in balancing the people, process, and technology impacts of adopting Generative AI tools and services. We can help determine the right policy and acceptable use guidelines for your organization. We can help you design experiences for customers, employees, citizens, and stakeholders of all sorts that could be augmented with Generative AI. Finally, we can help you build solutions that have the right guardrails to unlock the power of this technology ethically and responsibly in your business processes and applications. We can help you with the “where, what, and how.”

I think back to that same feeling of the world being different when I played with my first PC, downloaded my first browser, and made my first social post. The world feels different now. Each of those revolutions had a few things in common. First, they weren’t the overnight successes they appeared to be. Second, they combined existing technology with new ideas to change a paradigm. Lastly, they each made something that was prohibitively difficult for most people much more accessible. The PC took chip manufacturing and the advent of an Operating System to make computing more approachable. The web combined a more open networking scheme and content standards to make gathering information more approachable. The social web took improving programming frameworks and network speeds and made sharing and participating in the Internet more approachable. Generative AI is taking the power of advanced algorithmic techniques and web-scale cloud computing to make the power of AI more approachable. To my eyes, we’re standing at another fundamental shift in how we think about and live with a new technology. More and more people are going to get access to powerful tools, and the rate of improvement of those tools will benefit from exponential growth and improvement. I can't wait to see where this takes us.