In today’s episode, I discuss some of the most overlooked use cases for AI in business today. Rather than just generation, these powerful models can audit documents, data, and processes to find anomalies, improve quality, and ensure compliance. I explore how AI can analyze financials, check privacy policies, and even perform sensitivity reads on content. Don’t miss this insightful look at tapping into AI’s comparative abilities.
Can’t see anything? Watch it on YouTube here.
Listen to the audio here:
- Take my new Generative AI course!
- Got a question for You Ask, I'll Answer? Submit it here!
- Subscribe to my weekly newsletter for more useful marketing tips.
- Subscribe to Inbox Insights, the Trust Insights newsletter for weekly fresh takes and data.
- Find older episodes of You Ask, I Answer on my YouTube channel.
- Need help with your company's data and analytics? Let me know!
- Join my free Slack group for marketers interested in analytics!
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for watching the video.
In today’s episode, Ashley asks, “In your opinion, what are some of the untapped use cases for AI in businesses today?”
So there’s a couple of different use cases that people aren’t paying enough attention to because they’re being distracted by the shiny objects, particularly with larger context models now. So for example, the GPT series from OpenAI can handle a context window of anywhere from 16,000 to 32,000 tokens. A token is essentially a three or four letter fragment of a word. So 16,000 tokens, or any number tokens, is basically 66% of that is words. So if you have 16,000 tokens, you have about 11,000 words. If you have 32,000 tokens, you have about 20,000 words to work with.
Most of our business documents don’t clock in much higher than that. If you look at Claude from Anthropic that has 100,000 token context window, which boils down to about 60,000 words. Most books don’t have that, at least in business books. And that gives you the ability to prompt these artificial intelligence pieces of software to do both transformative work.
And I think the one that’s really overlooked is auditing work. Now we’ve all done some basic auditing tasks with these large language models, we’ve had them do things like fix spelling or fix your grammar or reformat the text to be more aesthetically pleasing to be more readable.
So we don’t really think of these tools as auditing tools in the sense of let’s do have a large language model do analysis on it. Again, a tool like Claude or GPT-4 can do extensive analysis on large amounts of data. And it doesn’t just have to be plain text, it can be PDFs, it can be spreadsheets, it can be, you know, any machine readable text format.
Think about this, if you were to put in say, all of your income, of your income tax forms into a large language model, and say here is the known tax code. So the time maybe you could even feed in the tax code or the sections that apply to you. Find irregularities, find anomalies, find opportunities to save money, right?
That auditing capability is something that large language models are capable of doing. But most people don’t think to do that. You can take, for example, your checkbook register from your bank, you can turn that into a CSV file, hand it to a large language model, you want to make sure the privacy settings are set so that they’re not recording your data. And then say, identify where I’m wasting money every single month. And it can look at your accounting data and say, “Okay, here are some possible candidates for things that don’t seem to make a whole lot of sense.”
These tools are very good at auditing in the sense of looking for fraud. Hey, here’s, again, a list of customer purchases. And you can say here are the ones that seem a little anomalous, you know, validate your sense of probability that this is a fraudulent transaction.
Auditing tools that are that are based in large language models are probably the most untapped opportunity these tools have to offer, because everyone’s so focused on them being generative and generative AI. It’s cool, right? You can make blog posts and news articles and newsletters and things that’s great, you should.
But they’re the mathematics underneath these models make them better at comparison than generation. So if you say here is my transactional data, compare it to known best practices for double entry bookkeeping. These tools can do that these tools can do that and say here are the anomalies, here are the things that don’t make sense.
All of these auditing capabilities things people are just not looking at nearly enough. And there’s tremendous value in that in helping us refine the work that we’ve already done, helping us identify problems, helping us elevate the quality of our work.
You know, these are essentially editors and proofreaders and inspectors and auditors who can look at our work independently and offer data driven opinions. Now, will they always be right? No. If it’s something that’s mission critical, please have a qualified professional, you know, do so. Look at it.
But these are some of the use cases, I think people are really missing out on they’re really just not paying enough attention and missing the benefits of some of these large language models. It’s a really good question. There’s a lot to explore. There’s a lot of different ways we can use these tools to to our benefit in a comparative sense rather than a generative sense, even though these models are capable of both.
So thanks for the question. We’ll talk to you soon. If you’d like this video, go ahead and hit that subscribe button.
You might also enjoy:
- Can Causation Exist Without Correlation? Yes!
- How To Determine Whether Something is a Trend
- How To Set Your Consulting Billing Rates and Fees
- How To Set Your Consulting Billing Rates and Fees
- Transforming People, Process, and Technology
Want to read more like this from Christopher Penn? Get updates here:
Get your copy of AI For Marketers