Generative AI -- what it is and why there's no rush to adopt it [Q&A]

Generative AI has been getting attention recently for its novelty, unique applications and potential impact on the business world.

But, like any new invention, there's some confusion around what it actually is and what it can do. We spoke to Scott Varho, chief evangelist of 3Pillar Global, who argues that companies shouldn't be rushing to adopt generative AI without considering their needs and potential value.

BN: We've all heard the buzzword, but exactly what is generative AI?

SV: Generative AI is a form of artificial intelligence that creates net new content, including text, images or speech. For example, ChatGPT, a model that interacts conversationally, is able to hold a dialogue with users. It's able to answer follow-up questions, admit mistakes, challenge users and identify inappropriate requests.

From simply inputting a quick phrase, generative AI like ChatGPT can create entire blog posts, poetry, song lyrics, code, artwork and more. Before the creation of generative AI, AI and machine learning (ML) could only analyze or act on existing data.

BN: What benefits can this offer to businesses?

SV: In addition to creating new content, generative AI shows potential for:

• Data generation: create large amounts of synthetic data or data about data, which can train other machine learning models or test new products and services.
• Data synthesis: plowing through large data sets to highlight potentially valuable patterns
• Personalization: personalized experiences and/or content for users within a product experience or a digitally enabled service.
• Automation: automate repetitive tasks, such as data entry or image annotation, which can save businesses time and money.

Although generative AI could significantly impact the business world, it's important to note that the specific benefits will vary depending on the business, industry and how the technology is applied. One thing I learned about the psychology of humans and data is that they are loath to make consequential decisions based on data they aren't sure is accurate or tells the full picture. This natural (and healthy) skepticism is increased when they don’t understand how the data was generated.

BN: What are the key considerations when looking at adopting the technology?

SV: Like any shiny new invention, companies shouldn't rush to adopt generative AI without considering the company’s true needs and potential value. New inventions get buzz and attention, but true innovation is doing or using something in a new way that creates value. Something that's been around for a long time can be even more innovative (generate greater value) than something novel. For example, instead of making product recommendations with an invention like ML, building a decision tree is a faster alternative to drive product recommendations -- and it's cheaper to maintain.

Simply put, companies can't embrace new technology for technology's sake. When considering whether or not to use generative AI, businesses need to consider their target market's needs to find the 'why' behind the technology and test the application in a lean way to understand its real value potential and limitations. Value should be the north star of building digital products that have a business impact and advance digital transformation.

BN: Which industries are likely to benefit most from generative AI?

SV: Many fields could benefit from generative AI positively because it can quickly create new content and ideas. For example, creative industries can use generative AI as a jumping-off point for works of art, music and literature. Marketing and advertising can use its content-generating capabilities for collateral and campaigns. Generative AI could also assist in developing new ideas and hypotheses, useful in research fields.

With caution, generative AI could influence product design and development. However, true innovation requires collaboration from craftspeople, who ultimately have the skills to make highly strategic business decisions. Good judgment and open discussions about trade-offs are key ingredients for delivering product value in the short and long terms, which AI in any form isn't capable of doing.

Ultimately, generative AI should assist humans and make them more efficient, not replace them altogether. It should eliminate repetitive, tedious tasks, but not take over critical thinking, insights, and decision-making. Before it does anything, though, AI should be carefully considered and tested for its value in context.

BN: How far away do you think we are from widespread adoption?

SV: This is a 'crystal ball' question and is heavily debated. If we look at another technology that has arrived, but adoption has been slower than expected, we can look at autonomous vehicles. The technology is now several years old, but the process of perfecting it and the efforts to take on the necessary legal and financial aspects (i.e. insurance claims for accidents potentially caused by an autonomous system) have taken longer. We will see two forces play out in the adoption: a rush to claim some of the hype while the real work of delivering value to customers continues.

I have spoken to leaders of businesses that have 'AI-powered products' customers are actively paying for. When asked how much they’re getting out of the AI portion of those products, they admitted only very little. The technology is still nascent, requires a ton of training and only does what it's designed to do. Still, ChatGPT has shown that it can provide an alternative experience to searching the web for knowledge, which is a threat that Google takes very seriously.

Meanwhile, AI-generated content of any kind that is trained based on materials authored by humans with rights will be challenged when it comes to attribution and royalties. Given these factors, I'd say we’ll see a lot of claims to be using AI (leveraging the hype) while actual adoption remains peripheral or complementary to the core of those products and services. Those who race ahead and figure out how to apply these technologies to generate value will likely be acquisition targets for their larger market incumbents.

If I had to make a prediction, the move towards widespread adoption is still five to seven years away. One of the primary barriers to adoption is talent at the execution and executive levels of organizations with a rich understanding of the benefits and limitations of the technology.

Photo credit: Peshkova / Shutterstock