Artificial intelligence: Small business

[00:00:00] Sanjay Parekh: Welcome to the Side Hustle to Small Business Podcast, brought to you by Hiscox. I'm your host, Sanjay Parekh. As a serial entrepreneur, I'm passionate about sharing the stories of other entrepreneurs. On this show, we share perspectives on all sides of the side hustle to small business journey — the challenges, triumphs, and everything in between. 

This season, we've met countless entrepreneurs who think outside the box to create the businesses of their dreams, but they aren't doing it alone. As the business world changes, so do the tools that we use to grow our business ventures. In today's episode, we are investigating how small businesses are using artificial intelligence, or AI, and if this tool is causing more harm than good for small business owners. 

Artificial intelligence — it seems to be the only thing people are talking about these days. From ChatGPT creating travel itineraries based on the prompt of a few sentences, to audio engineers mixing new music with well-known artists' voices, AI is impacting all industries and it's happening really fast, so fast that people are having trouble keeping up with it. 

In a survey by Business Intelligence Company Morning Consult of over 10,000 US adults, it was found that 10% of those surveyed found generative AI output very trustworthy, 11% found generative AI not at all trustworthy, and close to 80% of those surveyed are undecided at this point. With the technology so new and developing so rapidly, it can be hard to get a grasp on if it's ultimately a good thing or not. Is AI going to steal our jobs? Will it make select small businesses obsolete? 

Today we're going to speak to three individuals who work alongside AI, whether that be directly working on AI or incorporating AI into an already existing workflow. But for now, let's set a base and figure out exactly what artificial intelligence is and where did it come from. To start off, I want to introduce you to Chris Miller. Chris is a Seattle-based software engineer for Microsoft. Every day he works with emerging technologies such as artificial intelligence. 

[00:02:28] Christopher Miller: The AI systems I interact with include the new variety of GPT models. So, that's your ones like the ChatGPT that is publicly accessible as well as some internal variations on GPT models. And that is mostly within its integration into products within the sphere of AI itself. That includes things such as recommendation engines and stuff like that. And I will include things such as obviously like your Googles, your YouTubes, and any other search algorithms. 

[00:02:58] Sanjay Parekh: We asked Chris to help us understand the history of AI and we were surprised. Although it seems like artificial intelligence is something that has really made a name for itself this year, it's actually been around for decades. 

[00:03:10] Christopher Miller: So, what's interesting in a lot of the discussion of artificial intelligence within the media and stuff like that recently is AI in the workplace and just AI in general is not a new thing. We have had artificial intelligence in the computer science sphere, so this is now getting into like the computer science sphere of stuff, basically since around the 1960s. Artificial intelligence is very vague, so like me mentioning like search algorithms and like recommendation algorithms, technically that falls under artificial intelligence. Basically, there's a couple different spheres that it can fall under, which just basically AI in itself is just at its core, systems that can take a bunch of inputs and then produce some sort of output, and then usually those outputs are a bit nuanced in what they do, which kind of gives the semblance of intelligence. 

[00:03:56] Sanjay Parekh: Artificial intelligence has a much larger history than most of us know. Think about tools like Grammarly or even Google search engine. These tools are run using AI and have been around much longer than this year alone. Not only do these tools have a vast history, there's also variations in the types of AI that are available to the public. Today we're going to focus on two different types, generative and interpretive. 

Our next speaker is Dr. Angela Murphy, the Vice President of Business Development for Photon Commerce, an AI platform that helps users process payments, invoices, contracts, and receipts. The company has processed over $23 billion in payments thanks to artificial intelligence, superhuman speed and accuracy, as stated on their website. Photon Commerce is based in San Francisco, California, and Dr. Murphy is based in Kansas City. Here's Dr. Murphy to explain the differences between generative and interpretive AI.

[00:04:57] Dr. Angela Murphy: So, generative AI is that ChatGPT format where a user is inputting information and then the AI is generating a response back to that user based on the information that the user put in. Interpretive AI is artificial intelligence that reads and interprets information in a particular source. And then pulls that information and translates it into a different system of record. 

[00:05:21] Sanjay Parekh: With this information, let's look at the downsides of incorporating AI into the small business workflow before we get into the potential perks of the tool. There are a few reasons why artificial intelligence could be harmful to small businesses. But first, we want to talk about whether or not we should be trusting the output of these systems. In the survey by Morning Consult we referenced earlier in this episode, only 10% of American surveyed found generative AI outputs to be trustworthy. We asked Chris to explain why some might be cautious when using these systems.

[00:05:56] Christopher Miller: How these models are trained is, they're basically just given loads of information, just loads of text, and basically these models don't actually know right from wrong, real from not real. All they know is how to sound correct. So, when you give it a sentence, it's going to give you something that sounds correct and sounds contextually correct, but it can be complete nonsense. And that's what we see a lot on what the phrase that's used a bunch for is, the models hallucinate, which basically they just invent information. That's a big issue in terms of it being a cultural phenomenon. We now have people trusting the information that it produces when you should not trust it.

[00:06:36] Sanjay Parekh: This isn't the only downside to relying heavily on AI to produce content for you. Artificial intelligence works on the basis of inputted information from humans, and with humans comes bias. When these models are trained using sources from the internet, it doesn't know right from wrong, which can be dangerous.

[00:06:57] Christopher Miller: This also gets into very interesting philosophical arguments about how do you know what is true? This is a lot in terms of humans, we're bad at figuring out what new sources are correct. How on earth can we trust an AI to scrape data from it and then it also to know if it's correct or not? So, because of that, these AI systems are basically pre-trained with the same biases that Western Society has. And so, that is a huge hurdle to tackle in terms of producing systems that are equitable.

[00:07:25] Sanjay Parekh: It may seem like from these two examples that artificial intelligence is all doom and gloom, but luckily that's not true. In fact, Chris says that as AI evolves, so do the ways in which we are able to train it. 

[00:07:38] Christopher Miller: People are figuring out, you can basically train these models to, instead of making it up, know where to go look for the correct information. And so that's what we're seeing with a bunch of the tools and stuff like that coming out of Open AI where I was talking about previously. Like for example, if it doesn't know the answer, it could go look something up on Wolfram Alpha for example, which is a massive knowledge base, and it could be more informed for that and when it finally makes it the response for you. 

[00:08:05] Sanjay Parekh: Let's now look at it from a small business lens. We asked Chris what he thought the biggest threat to small businesses is when working with AI.

[00:08:15] Christopher Miller: I think the biggest threat to small businesses with AI is, owners under-selling the value of human work. 

[00:08:27] Sanjay Parekh: If tools like ChatGPT can produce reports, marketing material, and even provide customer service in a matter of seconds, where does that leave humans? What is stopping small businesses from underselling the value of human work? We asked Chris. 

[00:08:42] Christopher Miller: There's basically like a bunch of memes going around when like GPTs started on being on the rise and stuff like that of oh, replacing my whole workforce with AI, or, oh, replacing my CEO with AI. These systems aren't good enough to operate on their own. I don't know if they'll ever be, because as humans we have a lot. While we don't have the same knowledge base as these AI systems, because the AI systems are basically trained on the entirety of the human, the reported human existence, what we do have is societal nuance. And when we're building things and writing things and designing things for human consumption, we need that human nuance. 

Because we can design systems to sound like and act like humans all we want, but as humans, we're really good at picking out what is real and what isn't. That has an asterisk of sometimes we're really not, given things like fake news, etc., but like these mean when you see like an animation of a human and then you can immediately get the uncanny valley. With certain things we're trained on, and while we're not perfect with text because the text systems are also getting quite good. What we do know is, for example, if you're corresponding with a specific person, you know more about that person than this AI system will because it's a very, in the now moment, this is a person, like it's very short term. It wouldn't be in the system's knowledge base, so you can make calls that the AI wouldn't know. And sure, that can help you like inform, for example, your prompts to it, where it's like if you know this person likes things more casual in conversation or more professional, you can embed that into your prompts. But it's still going to require you to have that input and you to understand that stuff. 

[00:10:22] Sanjay Parekh: Chris is right. You have to remember that artificial intelligence is just that: artificial. It doesn't understand social nuance, nor is it able to comprehend the vibe of a company or the people it's assisting. 

Before we move onto looking at the positives and incorporating AI into the small business workflow, we want to introduce you to the final guest of today's episode, Raj Choudhury. Raj is the CEO of a company called Alloy, a public relations and customer experience company based in Atlanta, Georgia. For the last year or so, Alloy has been incorporating artificial intelligence into its workflow to help with things like creating interview questions and researching. Raj says that the company doesn't use AI with the intention of replacing their writers, but with the intention of adding another tool to their repertoire. Raj's biggest issue as he's been incorporating AI into the workflow of the company is retraining his current employees on how to properly use these tools. Raj says that because AI is moving so fast, schools can't keep up with the demand for knowledge, meaning there's no classes teaching this. Here's Raj.

[00:11:29] Raj Choudhury: In the late ‘90s and early 2000s, there were no universities teaching web design, just like there's no real universities teaching how to use AI workflow products and tools. So, there's a lag in standards and so forth. But I'd say that the talent coming through, today, the next generation are all going to natively know how to use a ChatGPT or Bard, pick your flavor of whatever it may be. The harder thing is retraining current folks to embrace it and not think's is something that's going to affect them. It does affect them if they don't get on board, so I think it's very much an evolution of any other product. If you said to a 23-year-old coming out of college, gosh, we should make sure you're trained on how to use Gmail. They'll laugh at you. Okay? So, of course you get these things become normal day to day, right? Word and spell check and Grammarly and all those various things. And by the way, calculators, right, at some point were a threat to schools and everything else, right? Excel and Google sheets and slides and so forth. It's an evolution of how we work. And I think the more a workplace can embrace those things, the more productive and the smarter our employees can be for our clients. 

[00:13:00] Sanjay Parekh: As the CEO of Alloy, Raj seems optimistic in the use of AI in the workplace, and while we have highlighted some of the cons of incorporating AI into the workplace, we wonder, is there a chance that the benefits of these systems outweigh the cons? After all these tools are extremely powerful and capable of more and more every day. 

Next, we are going to explore how these systems can work in a small business and the steps as a small business owner you should be taking to ensure their success. 

If we've learned anything so far in this episode, it's that maybe AI shouldn't be used as the end all be all for your small business. But that doesn't mean you shouldn't incorporate it into your workflow at all. In fact, as you will hear shortly, our guests today are all relatively optimistic about the tool, if it's used in the proper way, which a lot of organizations are already doing, which is, as Chris says, treating it as a coworker as opposed to the boss.

[00:14:04] Christopher Miller: For the vast majority of people and the stuff that they do with it, these models are definitely, they act in almost like a co-pilot capacity, like a coworker capacity, where you can ask the questions, it can get you in the right direction, and then you fix up what it does.

[00:14:18] Sanjay Parekh: By using AI in this capacity, you can still insert the human nuance into the content as well as double checking the work and fixing errors and not relying fully on the output of the system. This will really help speed up previously mundane manual tasks. For Dr. Murphy and Photon Commerce, this means better understanding your cash flow. 

[00:14:40] Dr. Angela Murphy: AI is going to give people the power to offload manual tasks and to make more strategic decisions with their time in their current roles. So, rather than spending, let's say, 10 hours a month or two working days a month hand-keying information, artificial intelligence is going to eliminate that particular task. But we're still going to need the humans in that workflow because this is where the generative piece comes in. You have artificial intelligence and interpretive AI pulling the information from paper, from checks, from invoices, from receipts, and it's going directly into the accounting system. But your accounting team still needs to look up, did we get paid for this invoice? Where's the check? Do the line items match? Does this match our contract terms? Like we negotiated this to be $3 per piece. And we charge them $3.25. We need to refund them. And so, AI in general for finance teams, for corporates, enterprises, banks, financial services, is going to help people better understand their cash flow. And that's super important in the supply chain. And for anybody that's in B2B commerce.

[00:15:47] Sanjay Parekh: Not only can AI eliminate mundane accounting work, but it can also eliminate mundane tasks when content creating, such as first draft and ensuring the nuance of the content is correct. Here's Raj on how once an employee grieves the loss of this work, it can better help the end result.

[00:16:04] Raj Choudhury: Maybe 30, 40% of what I do, but maybe that stuff was mundane, is now being replaced, and you have to grieve that and let it go. And once you do and you use it just like any other tool, you can start enhancing what you do. And the beauty of our people is really the ability to be creatives, right? To be storytellers, to have ideas. And that has never really been a massive threat for me from an AI standpoint. I think it can enhance those elements, but it's not going to come up with original thinking, just like a human can interpret something and create an original idea, but you can certainly use tools to speed up that process. 

We use a product called Writer that replaced pretty much a first-round editing of content creation. So, if you think about it, we work on various technology brands with different nuances and the personas they pursue, personas that they have tone and voice differs from one brand to the other. So, we can basically train, let's take X, Y, Z; A client versus B client and say, this is the voice and tone, and these are the specific keywords, or these are the things and the narrative we're trying to push through versus brand B and do the same thing. And then we can have any of our content creators create content and then run it through that model to then standardize a voice and tone instead of human editing to basically try and create a standardized voice and tone. So, that quickly sets the standard for the voice and tone from brand to brand to brand. So, we can have 10 different content creators, 20 different content creators creating content. The model then re standardizes it, then we bring it back to the writer to tweak it, right? And that allows us to basically have writers that can generate stories without having to worry about the nuances of specific words for each different brand that we're representing.

[00:18:06] Sanjay Parekh: And while an employee might grieve the loss of the work initially, Chris says that the systems that companies like Microsoft are producing are designed to take some of the mental load off on an employee's plate. 

[00:18:16] Christopher Miller: We're trying to produce systems that make things more convenient and take some of the mental and workload off of your everyday worker. It is very sci-fi the idea of you have a robot helper basically, and like in a lot of ways that's what people see GPT being, but yeah, I think it's yet to be seen its full impact, especially in the terms of small business. But in the short term, I could definitely see it's going to help people draft stuff quicker, whether it's emails, it's going to be able to help, be able to get through your inbox quicker. It's going to be able to help you write Excel formulas. It's going to be able to help you come up with ideas and pitches, lots of stuff in kind of that realm. Idealistically, it's going to save people time. AI tools help as it's like another search tool. 

[00:19:07] Raj Choudhury: I started using it just to enhance my workflows from everything from helping with interview questions, to research, to things like that. So, we're an agency that views it more so just like any other tool. Just like how we use Google Workspaces or how we use Microsoft's products or how we use different analytical products, or everything as simple as Grammarly, right? Things like that. Those are all based on tool sets that can help enhance workflows and how you use products on the day-to-day standpoint.

[00:19:41] Sanjay Parekh: Another practical use of the technology, as Raj has discovered: speeding up the hiring process of new employees.

[00:19:48] Raj Choudhury: I interview a lot of people to come into Alloy, right? And I use it. It's not that I can't come up with good interview questions, but I know that there's probably a better way of doing it and a faster way. So, I typically put in the job description into, I use Da Vinci quite a bit, so ChatGPT and Bard, but I'll put the job description, I'll put the values of my company, the values of Alloy, a little description of who Alloy is and the candidate's resume. And say, generate me 10 questions based on that. I've given it context, given it the reason, the values of the company, the things that I'm more interested in knowing and it's going to generate me 10 questions and five of those questions I’d have asked anyway. But the other five may be things that maybe I wouldn't have phrased like that but is better and maybe I'll take two or three of those and use those as well. So, it's just an enhancement. It doesn't replace who I am in asking questions. It just makes it better. Yeah, and I think any business can benefit from that to just make our businesses better. 

[00:20:55] Sanjay Parekh: Using artificial intelligence as a tool as opposed to an employee seems to be the answer on how to use these systems correctly. Human workers keep their jobs and use AI as a tool to enhance their work, not replace it.

[00:21:09] Dr. Angela Murphy: One of the concerns that people bring to me quite often lately is, is AI going to steal my job? Or are the robots coming for us all, right? And I have to gently remind everyone that this isn't Terminator, like Arnold Schwarzenegger is not going to show up. It's fine. The other thing that I will say is that AI technology, RPA, those things are allowing companies to scale more quickly. So, when you're an SMB, for the most part, you want to scale your business, but it's really challenging to do because you can't always hire the headcount you need to grow and to fill in those gaps as you're trying to go on that growth trajectory. So, now that we have AI really at the forefront of the conversation, SMBs are more empowered to make better decisions about personnel. And to marry that with technology to allow them to grow more efficiently. 

[00:22:02] Sanjay Parekh: So, while AI is likely not going to steal your job, it can help small businesses grow their company without having to hire a full team for every department.

[00:22:12] Dr. Angela Murphy: So, especially like smaller teams who maybe don't have a full marketing team, where they have one person that's doing marketing, but also maybe wearing a couple of different hats. What I'm seeing is that people are using within their small business, the generative AI to simplify some of their writing tasks or to simplify their scheduling. Or let's say you're trying to figure out the right workflow to do LinkedIn or social media posts. You can literally type that into a ChatGPT or to an Open AI, and it can give you, hey, this is the schedule that you need to follow, and here's some sample posts. 

[00:22:53] Sanjay Parekh: So, AI can help with a multitude of things within your small business. It can take away some of the long, mundane tasks that would normally take days, and allow your employees to focus on the more important issues at hand. It can standardize your work, allowing multiple employees to work on the same task without having to worry about nuances of specific words they're using, and it can help scale your small business even when you don't have the budget to hire a full team in different departments. To wrap up today's episode, we want to learn what's next for small businesses and artificial intelligence. And what does the future look like as the technology continues to evolve?

At the end of the day when used properly, AI is an incredible tool, as we've just touched on. However, not everyone or every company is sold on it. We asked Raj about his thoughts on companies who are hesitant to jump on the AI train. 

[00:23:52] Raj Choudhury: There's a number of agencies who are rebelling against using AI products, right? Because it takes that, they feel it diminishes the creativity. And I'd argue that it does if you let it. If you replace the function of, listen, you shouldn't go into ChatGPT and ask it to write a press release, right? It can. It can create a long form story, but it's not going to be that great because you haven't given it context and meaning and purpose and a storyline to really follow through.

It can enhance elements of it. So, I think agencies who are holding back because of this notion it's like saying back in the day, hey, let's mail people things instead of emailing people. It's that kind of element. So, I think workplaces, whether small businesses, agencies, whatever it may be, need to think about how they can enhance things.

[00:24:54] Sanjay Parekh: And Dr. Murphy's final thoughts? Being fearful is going to limit us as a society. 

[00:24:59] Dr. Angela Murphy: I think, for the general public and also practitioners who are trying to understand how artificial intelligence is going to affect your life and affect your work. I would encourage too not be so fearful. Artificial intelligence technology is going to be to work in how people engage with the world now as cars were to horses. That is the step that we are talking about in terms of how this technology will impact everyone. We have just started scratching the surface. And being afraid of it is not going to help us figure out the best ways to use it and to move us forward as a society.

[00:25:45] Sanjay Parekh: We've only just covered the tip of the iceberg when talking about artificial intelligence. And while our guests all seem optimistic about the uses of AI in small business, we encourage you to explore the topic further to see if it's a good fit for your organization.

Thank you for listening to this episode of The Side Hustle of the Small Business Podcast. To learn more about Hiscox Small Business Insurance, visit the Hiscox blog at www.hiscox.com/blog.