How AI May Change EntrepreneurshipFrom coming up with an idea to creating a marketing plan, artificial intelligence adds a whole new element to starting a business
Artificial intelligence is about to change how entrepreneurs start companies—and the odds of making them successful.
Systems such as OpenAI’s ChatGPT, Microsoft’s Bing and Google’s Bard can assist in almost every step of devising a startup, from coming up with an idea and testing it to performing consumer research and creating a marketing plan. They do those jobs faster and more efficiently than an individual could, and in ways that a human isn’t capable of carrying out.
“In 12 months, it’s going to be a very different way of starting a company,” says Steve Blank, a longtime entrepreneur who helped launch tech companies including MIPS and Convergent Technologies, and is an investor in an AI cancer-diagnosis company.
Below are some of the ways AI can aid entrepreneurs—as well as some caveats about the technology’s limitations.
Generating the bright idea
Finding an unmet need is the first step in creating a startup. It takes inspiration and intuition, careful research and the balancing of risk and reward. And it may turn out that AI can do this faster and more efficiently than any human ever could.
Such systems “could eventually become some of the best identifiers of unmet human needs in history,” says David Schonthal, director of entrepreneurship programs at Northwestern University’s Kellogg School of Management.
That is because AI can absorb vast amounts of information and sort through it to figure out what people might want in the marketplace. “AI reads a bunch of data to identify patterns in that data, it acts upon what it senses and then it learns based on what it puts out in the world,” Schonthal says.
That is what entrepreneurs do, he says. But entrepreneurs “are limited by the constraints of their own capacity, their own reach and their biases,” he says. “AIs have an exponentially wider set of data to learn from.”
For instance, an entrepreneur could ask an AI system to identify problems consumers face in using certain products or services, and then have it suggest solutions, says Ethan Mollick, an associate professor who teaches innovation and entrepreneurship at the University of Pennsylvania’s Wharton School. The entrepreneur could add limiting factors, such as the cost of each proposed solution. The system may come up with dozens of ideas and variations that an entrepreneur may not have considered, Mollick says.
AI can also help focus an idea. AI systems can be asked questions and then respond: For instance, entrepreneurs can ask the AI to refine its answers, such as to consider a different variable in a startup proposal like whether a product would be popular abroad or whether it could be priced higher. Or entrepreneurs can ask an AI to give an example of a similar product that thrived—or flopped.
“It’s like having the best business-school coach you could have,” says Blank, the entrepreneur.
Surveying the competition
Sure, entrepreneurs may think they have a smart idea for a startup. But perhaps a dozen people already have thought of it. The vast databases that AI systems draw from can provide a bigger research reach than typical internet searches.
Using AI, entrepreneurs can see what has been reported about a business concept in technical reports, financial filings and news articles. They can scan marketing and consumer surveys and check what has been said on social media about related startups. They can see what types of startups were failures and which have thrived.
Testing the concept
Even if entrepreneurs think their startup idea is smart and investment-worthy, they need to test their hunch. AI could help by playing the part of a customer based on a digital persona founders give it.
For example, if a company concept involves the dental industry, entrepreneurs could tell an AI system to pretend it is a dentist with 20 years of experience who runs their own practice. Then the entrepreneur would ask questions about what constraints the “dentist” faces in doing the job and whether the proposed product or service would help.
“You can have a reasonable interview with it,” Mollick says of the dentist persona. “You will not get the same stuff you get out of a human. But if it helps you refine an idea, it’s really valuable.”
The negligible cost and great speed of AI testing is a plus, says Blank. “I could run experiments against hundreds of thousands of different personas simultaneously and I could create websites with artificial products that I could test,” he says.
Moreover, an AI-based system could boost the use of “A/B” testing, in which two or more versions of a website or product promotion are put online to see which generates greater or more positive response. AI could create numerous versions of the tests, promote them through social media and targeted advertising, and then constantly refine the tests based on the responses, says Northwestern’s Schonthal.
Creating a business pitch
AI could also help devise and critique the business pitches that entrepreneurs send to venture capitalists. And it could help an entrepreneur figure out which VC firms are funding what types of startups—and which they are rejecting.
Mollick had one of his classes create startup proposals, then had an AI system—instructed to act like a venture capitalist—critique them. The AI system told the students whether it thought the proposals were unique, potentially profitable, addressed a market need and whether they adequately accounted for factors such as competition and the ability to scale up.
“I had a real VC critiquing them and then the AI VC critiquing them,” Mollick says, and the real VC thought the AI critiques were on the mark.
Promoting a new company
Once a company is funded, it needs to get promoted. Mollick ran an experiment to see how AI could assist in promoting an educational game. “What it accomplished was superhuman,” he says.
First, he asked Bing to find detailed information about the game, an online tool created by Wharton for teaching leadership to business students. Next, he told Bing to pretend it was a marketing expert and come up with a proposed promotional campaign. It produced a credible plan, including identifying the target audience and creating four sample emails touting the game to business educators.
In just 30 minutes, the AI systems that he used “did market research, created a positioning document, wrote an email campaign, created a website [for the product], created a logo and ‘hero shot’ [large centerpiece] graphic, made a social-media campaign for multiple platforms, and scripted and created a video,” Mollick said in a blog post describing the experiment.
Without AI tools, it would have taken him “many hours, maybe days of work,” he wrote.
Not a magic potion
Yet as with all users of AI, entrepreneurs need to be cautious. These systems are notorious for “hallucinations”—giving answers that sound authoritative but are misleading or even made up. What they spew out needs to be verified, experts say.
“I wouldn’t trust any numbers it’s giving you,” Mollick says.
AI presents another concern: It could come up with ideas that are “literal and obvious, which is a way to not necessarily be successful,” says Northwestern’s Schonthal. “What AIs will do is sense what people say they want, which is often different from what they truly desire.”
As a hypothetical example, an AI system could discover that sales of cheese are rising, and in response propose creating a new cheese-distribution system instead of “a truly innovative new business,” Schonthal says. “That abstract piece of the process is still a uniquely human capability.”
Moreover, most AI systems are “trained” on databases of things that happened in the past. “Good entrepreneurs are the ones that are thinking about what might happen in the future,” he says.