The cost of starting a business can vary greatly, and many successful ventures have begun with minimal initial capital. Two friends, Sal Aiello and Monica Powers, have turned a small investment of $185 (around? 15,000) into a lucrative business. They created a tool called DimeADozen, with the help of ChatGPT, to test business ideas. They sold the startup last month, raking in more than $66,000 in profits. “This proves that AI can print money,” Aiello tells CNBC Make It.
Aiello and Powers met at a virtual startup founder meetup event, where they began working on their side hustle together. They had no business plan or funding but could turn their idea into an AI-powered research tool in just four days, spending $185. The duo developed the tool to test their business ideas but soon realized that others could benefit from it, too. So, they built a website that allows entrepreneurs to share their ideas with the bot, and the software will return answers to potential investors, customers, and competitors. It’s the first AI tool explicitly designed to test out business ideas, and Aiello claims it works faster than Google searches and is more accurate. It also uses prompt engineering techniques to minimize the risk of “AI hallucinations” — where the chatbot gives plausible information but false information — a common problem with chatbots.
Besides providing business-related information, the platform provides more general help, such as explaining quantum mechanics, planning a fun day in New York City as a tourist, or writing computer code. It can also help solve math problems or give you the results of various experiments and studies. But it’s important to note that the bot could be better. It can make mistakes in its responses, including referencing nonexistent historical figures and books, inventing fake scientific data, or needing help to solve specific math problems. And with millions of people using the same AI model, hackers can exploit these vulnerabilities by injecting malicious data.
ChatGPT has a bug bounty program that rewards white-hat hackers who report vulnerabilities in the model to prevent this from happening. The company has hundreds of millions of users, so it has much to lose from a single flaw in its AI model. But even with the best security measures, no digital system is 100% secure, and hacking can occur through several means, such as data poisoning, which involves injecting malicious data into the AI model that affects all its future responses. This is why protecting your device with a password and using strong, unique passwords is essential. You should also change your passwords frequently and use a secure Wi-Fi connection.
