Entrepreneurship in the Era of Disinformation

EDITOR'S NOTE: Steve Blank has had a celebrated career as a serial entrepreneur and founder of the Lean Startup movement, which for the first time gave startups a methodology for success. He received the Lifetime Award for Entrepreneurial Excellence from the U.S. Association for Small Business and Entrepreneurship for the impact his work has had on entrepreneurship and entrepreneurial education. 

Today generative AI is disrupting the process of researching innovations and markets, so central to launching a successful business. Misinformation and disinformation – deliberately spreading wrong information to manipulate people – have made it more difficult to tell fact from fiction.

EIX recently asked Steve how entrepreneurs and innovators can navigate in a new world where information gleaned online is more unreliable.

About a year ago you wrote that AI will revolutionize the customer discovery phase of Lean Startup.  Since then we’ve seen that AI-generated images and information can be used to mislead and manipulate others.  “Disinformation” aided by AI has been a big concern. Has this changed your opinion of how AI research fits into the Lean Startup process? If so, how? If not, why not?

Any time you do market research you need to understand the source of the data. Any source that sounds authoritative could be disinformation, whether it's AI or anything else. Take Wikipedia. 95% of people use it as the ultimate source of truth. That's not AI, but we tend to lean on what appear to be authoritative sources. More recently, the volunteers who manage Wikipedia have had problems keeping up with vast amounts of AI-generated content, which can produce wrong information and have inaccurate or missing sources.

AI is going to create a whole new set of pseudo “authoritative” sources, much like Wikipedia, where the truth is not clear. Unlike Wikipedia, the problem of hallucinating footnotes and sources also becomes a danger.

Each layer of technology between us and the actual facts gets us closer to George Orwell’s  book “1984,” where the “Ministry of Truth” spreads lies and the “memory hole” erases all information that points to the truth.

When I'm doing any serious work, I check multiple sources and try to find out the source reference. I think sites like Snopes are going to become more and more important.

Will AI ever replace on-the-ground research with customers?

Remember, the Ministry of Truth in “1984” was really the Ministry of Propaganda. It was responsible for falsification of historical events and decided what the truth was.

The problem is that AI allows bad actors to become the Ministry of Truth. Right now, our equivalent to the Ministry of the Truth is Wikipedia and, in general, Google Search. As truth is shaped by these different forces, it’s going to become much harder to find the truth online. You will still need to see many things for yourself. 

When is it still a smart move to gather information by talking with customers instead of relying on research conducted with the help of Chat GPT?

AI makes it even more important to get a reference ground truth — meaning is what I'm reading/hearing/seeing, a hallucination or is this really backed up with some real data?

Consider this: During the pandemic, we discovered that you didn't need to get out of the building for 100 percent of discovery interviews because video conferencing allowed you to reach 10 or 20 times more people.

You traded off breadth versus depth of interviews. But in hindsight, not meeting people in person actually makes a good chunk of your conclusion suspect because you lack ground truth to test against – something that you’ve observed firsthand or from sources that are undeniably trustworthy.

AI makes this 100 times worse. It's like doing 100% of customer discovery on video. You miss all the subtle clues that human interactions give you about place, person, etc. I think we're going to see the same problem with complete reliance on AI.  

In your previous article about AI for us, you said you’ve been telling students to use AI to help them build business models and do discovery. Has that advice changed?

I'm telling my students, the problem in a transition of a disruptive technology, is that if you blink, you miss another useful tool. 

Meaning that, when we went from client server to the Internet, you literally had to keep up daily with what the new apps and services available. I think we're having the same problem as the onslaught of tools for students and for new companies hasn't slowed down.

The biggest pitfall will be getting stuck in a tool that you loved last year.  I will guarantee you, there is one 10 times better this year.

Parting Thought

The big idea is that the Lean Startup was not an incremental change, but a sea-change. It’s a radical disconnect between how we used to think about startups in the 20th century and how we now take for granted how we think of them in the 21st: the whole methodology, search versus execution, different tools, business models versus business plans, etc.

AI has the potential of being another disruptive break to accelerate startup velocity and to reduce early stage startup failures at that same scale. The dust hasn't even begun to settle with the number of tools that are going to be available. But AI will be as important as the lean methodology and providing a tool set and language for entrepreneurs.

There’s going to be a set of have-nots, people who get left behind trying to compete.

And when someone has a tool that allows you to operate 10 times faster, you don't want to be the person with a shovel when someone else has a bulldozer.