VIDEO TRANSCRIPT: There’s a lot of talk now, the biggest hype, really, in the industry of marketing, is using artificial intelligence, these large language models, which have tons of data that you’re able to query and get information back. A lot of people are using it for generating images, video, blog, audio, all these things can be created. But what we found is, when you take a business approach to it, it’s really no different than the software development life cycle, or what was called Change Management and Digital Transformation as the web had come on. You really have to look at this stuff through the lens of business. What kind of changes are you going to make? 

The best place that we’ve seen it is in extraction and summarization, being able to load huge amounts of data into these models and have it give you insight. Imagine, for example, if you have a product where there’s 20 product manuals, each of them is 1000 pages. To be able to instantly get answers and find out how to solve problems without having to read through all those manuals, that’s a huge advantage, and can put you way ahead in writing relevant content and stuff that works. So as you’re considering which AI tools to use, look at it through the lens of business, figure out what processes you want to automate. Where do you think you can get more advantage out of AI? 

Then go ahead and load the data you know that you have into it. There’s one side point on this too we have to keep in mind. In general, if you’re using a paid service, it’s more likely that your data is not being used to train the model or being used by other users. So you want to gravitate towards paid services if you’re concerned about privacy, data security, especially for a smaller business. Google’s Notebook LM allows you to upload your data sources and then do queries against it, to have it tell you what’s in there and where it’s going. The thing with this is you get a paid version so you know your data is not being used by anyone else, and it is limited to the data that you upload. 

One problem we see with a lot of other large language models, with chatgpt and those types of things, is that they can hallucinate. They will actually give you incorrect answers, or just make answers up that are completely irrelevant because there’s no obvious answer to the questions that you’ve given it. The thing to keep in mind is all of these large language models are the summation of all the information on the internet. So if you’re talking about a topic that’s pretty well mature and everybody understands how it works, you’ll tend to get really solid answers. If you’re talking about something that’s hotly contested and has a lot of different viewpoints, then the answers that you get might be all over the place and might not be that accurate. 

Again, this is moving so quickly that I don’t even want to talk specifically about solutions, but I can say, if you check out the MFN network, they have an AI board there that talks about AI tools and solutions that startup founders have used, and it has the use cases.  It’s not just saying, “hey, this model’s the best.” It’s more of okay, if you want to do stuff with payroll, or if you’re working on your website, it talks about what tools have worked best for that and where they’ve seen success. So a resource like that is invaluable as you dive into these new tools that are always changing.

Are you working to find community and opportunity as a Massachusetts startup founder? Join MFN and get access to experts, networking opportunities, and connections within the statewide entrepreneur ecosystem.