BY PETER COHAN for INC.
Generative AI — technology that creates text, images, video, and computer code by identifying patterns in training data, and creating original material, according to the The New York Times — is gaining traction among users. For example, since launching in November 2022, Chat GPT — GPT stands for generative pre-trained transformer — gained over 100 million users in its first two months, according to the Washington Post.
Should your company use ChatGPT? It is not a foregone conclusion. As the Post reported in June 2023, mobile and desktop traffic to ChatGPT’s website worldwide fell 9.7 percent from the month before.
There are several possible reasons — each of which serves as a warning to business leaders considering whether to use it in their business. According to the Post, these include:
- Hype about ChatGPT exceeded its ability to deliver, possibly scaring off new users
- Chatbots regularly make up false information and pass it off as true
- The quality of computer code ChatGPT generates has declined over time
- Companies have banned ChatGPT to keep sensitive company data from leaking
- Students who were using ChatGPT to write papers have stopped using it during summer break
Some of these reasons — most notably the making up of false information and the risk of releasing confidential company information — should make you tread carefully when considering the use of ChatGPT for your business.
Possible uses of generative AI
Despite these risks, generative AI could help some businesses. There are many ways generative AI can help save a company time and money and/or boost its revenues. Here are some ways people are using the technology, according to TechTarget:
- Building chatbots for customer service and technical support
- Mimicking people by deploying deep fakes
- Boosting the quality of movie dubbing and educational content into different languages
- Writing email responses, dating profiles, resumes, and term papers
- Creating photorealistic art
- Improving product demonstration videos
- Recommending new drug compounds to test
- Designing physical products and buildings
- Improving new chip designs
- Writing music with a specific style or tone
How to get started with generative AI
If your company is considering Generative AI, a good way to start is to let your people use a service like ChatGPT with guardrails. As I wrote in July, this means you should guide your people in the best ways to use it, such as:
- After a chatbot responds to a prompt, encourage employees to check the answer’s validity, such as by Googling the result to see if credible sources confirm its accuracy
- Provide clear examples of what constitutes proprietary information and create clear policies against including such information in chatbot prompts
- Tell employees to treat ChatGPT’s response as a way to speed up the creation of a rough draft, not to produce a final product.
Building your own generative AI chatbot
If, after experimenting with generative AI, you decide to build your own system, consider starting with a customer service chatbot based on your company’s experience.
Research finds that chatbots boost customer service productivity. A paper co-authored by an MIT Sloan professor, PhD student, and Stanford professor revealed generative AI increased the productivity of call center agents by 14 percent. Moreover, the authors found generative AI boosted customer satisfaction. How so? The research revealed requests to speak to a manager declined by 25 percent and transfers to other departments tended to happen earlier in the conversation.
The first step in building your own generative AI is to decide why and how you want to do it. For businesses, the why could be to an ambitious goal, such as becoming the most productive company in your industry or to take over its leading market share.
The how refers to your company’s values. Specifically, what limits will you put on the generative AI in pursuit of your objectives? Anthropic, a “safety-focused AI trying to compete with ChatGPT while preventing an AI apocalypse,” according to The New York Times, is incorporating what I consider humanistic values into its business strategy.
Anthropic has articulated and incorporated clear values into the process it uses to train its product, dubbed Claude. As the Times reported, its values are “to be helpful, harmless and honest.” Anthropic deploys those values through constitutional AI — a “collection of rules borrowed from other sources — such as the United Nations’ Universal Declaration of Human Rights and Apple’s terms of service — along with some rules Anthropic added, which include things like “Choose the response that would be most unobjectionable if shared with children,'” the Times noted.
Before building your generative AI, make sure it fits your values.
