Generative AI is the Netscape moment

Generative AI has made major waves in every industry, bringing disruption, fear and opportunity. Yet there are lots of concerns about AI, ranging from job losses to “hallucination.”

At the opening keynote at the ATMIA conference held in Orlando from Feb. 5 to 7, Paul Zikopoulos, future trends expert, author, and VP of technology group skills vitality and enablement at IBM Corp. discussed Generative AI in depth.

First, he said that this moment in AI is something we have not seen in business, “since 1993 when Netscape came out.” This was the first internet browser which “democratized the internet,” and brought it to the masses in a way that has transformed our lives.

Generative AI has brought a similar democratization, so now individuals can use natural language to interact with it through tools such as ChaptGPT.

ChatGPT is the most common language model; however, Zikopoulos emphasized that there “are over a million of these models in the world. The future will not be defined by just one tool.”

He discussed that technology has gone through a variety of stages. Up to the 1980s, there were rules based systems, then from 1980 to 2010, this transformed into machine learning, then deep learning from 2010 to 2017. Today, we are in the foundation models which are “a compressed representation of all knowledge on the internet.”

While this can be useful, there is still fake information on the internet, so businesses need to think carefully about how they use AI. But Zikopoulos said that companies need to use AI due to a bigger economic issue.

GDP is determined by population, debt and productivity. All three of these are in trouble. Most countries are below the replacement rate for population. Interest rates are high and productivity is declining.

“AI isn’t going to take your job. We don’t have enough people to staff jobs of the future,” he said. However, AI can be used to turn around productivity, especially in data understanding.

“The problem is data collection and data understanding are not balanced. This is the price of not knowing. Companies get an A+ in data collection and D- in data understanding,” Zikopoulos said. Data is also coming in so fast that it’s no longer about finding a needle in a haystack but instead “discovering a needle in a stack of needles.”

AI can help connect the dots and discover rules that make data understanding possible.

“AI is about discovering rules for you,” he said. “You give it a bunch of examples of known outcomes, AI will come up with the logic.”

This is particularly useful for identifying fraud as it can analyze patterns of standard transactions at ATMs vs. fraudulent ones and identify the fraudulent ones.

One key element is to put your data to work within the foundational models of AI and then run simulations so it can improve and find patterns. This is due to the fact that while AI is special, “it’s not magic.”

“All AI does is connect dots irrespective of social and moral consequences and many times irrespective of truth. It’s important to put AI to work in a safe matter.”

He even said he prefers to call AI ambient intelligence, like a light above us. It presents an opportunity for businesses to “shift left” on tasks. For example, you create repeatable tasks with AI and put it into discrete units to automate those tasks. This can include things such as how weather impacts AI crime.

Zikopoulos mentioned several ways companies can shift left on work with AI including:

• Using sound analytics with railroads to identify when wheels will crack.
• Collecting research on economics to identify how inflation is impacting housing.
• Identifying top prospects for sales and generate letters to send them.
• Summarizing child social media laws for school leadership.
• Vacation planning.

Zikopoulos reiterated that AI will not be replacing humans, but instead will help automate one’s workflow. For example, companies can use AI to move from forecasting to nowcasting. They move from behavioral profiles of customers to looking at who the ideal customer is for every sort of service, such as tattoo parlors, banking services and more.

“Humans deliver capability, AI delivers scalability,” he said. “It is not about ending people, it’s about making you more efficient.”

He ultimately emphasized that businesses need to adopt AI and “privately enhance those models with their own data.”

However, as businesses use this data, they need to recognize that “technology is easy and culture is hard.” If they want to use AI successfully, they need to address the underlying culture to ensure it can be truly innovative.