Learn to leverage AI and ChatGPT to improve your day-to-day operations.
As both a founder and small business owner, I’ve seen firsthand how today’s rapidly evolving business landscape calls for innovative ways to gain a competitive edge. Artificial Intelligence (AI) has emerged as a transformative force that offers tremendous potential for businesses of all sizes — as long as we understand how to use it to our advantage. The advent of AI, including and especially the advancements in generative models like GPT-3 and GPT-4, has opened up exciting opportunities for small business owners to harness the power of AI and drive growth.
What is AI?
AI is defined as intelligence or the perception, synthesis, and inference of information by machines rather than humans or other animals. We expect humans and animals to have some level of intelligence, but machines, which are inanimate objects, aren’t supposed to, so their intelligence, which humans program, is artificial and due to our scientific advances.
Some examples of AI include:
- Speech recognition
- Computer vision
- Translation between natural languages
Some AI applications include:
- Advanced web search engines (e.g., Google Search)
- Recommendation systems (used by YouTube, Amazon, and Netflix)
- Understanding human speech (such as Siri and Alexa)
- Self-driving cars
- Generative or creative tools (ChatGPT and AI art)
- Automated decision-making
- Competing at the highest level in strategic game systems (such as chess and Go).
As you can see above, AI has tons of different functional capabilities. One of those, and the focus of this article, is generative AI, which refers to systems capable of generating text, images, video, 3D models, or other media in response to input of some kind. Examples include:
- Text to text (ChatGPT)
- Text to image (Canva)
- Image to image (Adobe PS Generative Fill)
- Text to video (DeepBrain AI)
Generative AI models work by learning the patterns and structure of their training data, then generate new data with similar characteristics based on what they’ve learned. GPT and all its variations are generative AI — meaning they learn based on inputs to spit out brand-new outputs. Those outputs include actual data and information (usually — we’ll get to that), but what they generate is unique and done by the technology itself.
Adobe Photoshop’s Generative Fill Tool
Adobe Photoshop released its version of generative AI last week in the form of its “generative fill” tool, and I think it’s the perfect way to visualize generative AI.
You provide the generative fill tool with input — the existing image — and based on what it “sees,” Photoshop generates the new content to match.
If you’ve heard of absolutely anything AI-related in the last six months, it’s ChatGPT. We’re going to take a step back before we talk about ChatGPT and touch first on the technology behind it.
GPT stands for generative pre-trained transformer.
Generative: capable of creating coherent original content
Pre-Trained: the model is first trained on a large quantity of text data, with the assumption that it will later be fine-tuned for a specific NLP (natural language processing) task, like language translation.
Transformer: neural network model that includes self-attention, or the ability to differentially weigh the significance of each input they get
So, GPT models are a type of LLM (large language model, meaning they are pre-trained on large sets of text) based on transformer architecture, able to generate novel, human-like content.
GPT models are not only pre-trained on the text and parameters inputted by their creators, but they also learn constantly from everything we, as users, input.
GPT technically refers to any model that falls underneath this definition but has most commonly been associated with OpenAI’s versions — they introduced the first one and have continuously been the best and most popular.
OpenAI, an American AI research laboratory, is responsible for most current GPT models. All the models we currently have are considered foundational models, which means they are trained on broad data at a scale so vast that it can be adapted to a wide range of downstream tasks — essentially, they are so massive and trained on so much data that they can be used for almost any generative tasks.
GPT-3 and GPT-4
GPT 3 is the first model we really heard of because it was the first one open to the public and is the basis for the current, free version of ChatGPT. GPT-3 is unimodal (it can only accept text input), trained on 175 billion parameters, and performs poorly on complex tasks. It is also prone to errors and hallucinations — essentially, making things up.
GPT-4 is the latest and greatest GPT technology and has some huge updates that have already marked a massive AI shift. GPT-4 can handle conversations more profoundly and answer more accurately and precisely. It is multimodal (it accepts both text and image input), trained on an undisclosed number of parameters, and has performance close to human level, even on complex tasks. It’s not perfect, but it is considered much more reliable and factual than GPT-3.
There is actually a GPT-3.5, a slightly improved version of GPT-3, that is currently underlying ChatGPT, but that model can be primarily lumped into the same category as GPT-3.
It is essential to understand GPT models because they are the basis for AI tools like ChatGPT and other generative AI tools and applications. They represent the actual tech behind the products and programs we’re starting to see pop up everywhere.
GPT-3 is currently open-source, which means anyone can harness it and use it to build their technology.
ChatGPT is by no means insignificant in the AI world, but it should be noted that it’s also not necessarily a shining example of all that AI is capable of. That being said, it has the capacity to make a big difference in your business and your life and is well worth understanding.
Fun fact: ChatGPT is the fastest-growing consumer application of all time, with over 100 million users in its first two months.
ChatGPT is super versatile — it can do everything from writing and debugging computer programs to writing business pitches, translating and summarizing text, and playing games. It is trained on 175 billion parameters but also learns from everyone who uses it. As of March, it’s able to run plugins, and pretty much anyone can develop and submit them — I know there are already plugins available from Expedia, OpenTable, Zapier, Shopify, Instacart, and Slack that help ChatGPT perform even more tasks. For example, you can now ask it to write you a weekly meal plan, create a grocery list based on it, and order your groceries.
Its drawbacks include limited knowledge of anything that has occurred since September 2021 and nonsensical or plausible-sounding but completely incorrect answers. Check out OpenAI’s playground if you’re looking for something more versatile than ChatGPT or want it to read URLs to grab extra information.
ChatGPT goes from a good resource to an absolutely spectacular resource once you know how to prompt it effectively. There are tons of lists and articles and Twitter threads now with tips and tricks for prompting, but my favorite is the PROMPTER method by SpaceCadet’s team.
You could ask ChatGPT to “describe a mission to Mars,”… or you could input:
P — You are ChatGPT, an AI Language Model with in-depth expertise in intergalactic space travel and exploration.
R — Imagine you have just been selected to travel to Mars, a planet you have spent years studying.
O — Describe the mission you have been selected for from the perspective of an astronaut, including key objectives and obstacles.
M — Use language consistent with trained astronauts and astronomers. Focus on specifics unique to travel to Mars: distance, atmosphere, and terrain.
P — Adopt the perspective of the astronaut, showcasing several emotions around the mission to Mars:
- Uncertainty before being selected for the mission
- Elation upon being selected
- The awareness of the danger and enormity of the journey ahead
T — Write the request in the form of a journal entry, adopting the perspective of the astronaut realizing they have been selected to make history.
E — For example: “Day 3: As I prepare for the journey to Mars, I am actually aware of the enormity of the task ahead. The distance between Earth and Mars is vast, and the journey will take months of travel through space. This is the adventure of a lifetime, and I’m ready to embrace it with all that I have.”
Then, once you get a great (albeit long-winded) answer, you can say:
R— This is great. Please tighten the journal entries, keeping them to a maximum of 280 characters. Also refrain from using too many technical terms to increase readability.
You don’t have to use the PROMPTER method every time, but understanding the fundamentals and the way in which it helps to maximize returns on your input will turn ChatGPT into a force for your business.
How to Train Your Chatbot
Right at this very moment in the history of the world and the internet, everybody wants a chatbot to answer all their questions and solve all their problems. You can have a chatbot be your girlfriend; you can make your weekly meal plan and order your groceries; you can replace your travel agent; you can have a free personal assistant; you have generative AI at your fingertips for the first time in history.
We at Kairos harnessed the hype by creating KairosGPT, a chatbot trained on The Psychology of Money by Morgan House (our favorite finance book!). Since OpenAI so kindly made GPT-3’s technology open source, all it took to create this was a bit of coding by our engineering intern and an API key. Chatbots are a terrific way to gain attention. They can even be an MVP or a hook, immediately giving your customers the information and answers they want while drawing them into your business and proving your ability to help them.
Check out Botsonic to build your chatbot — no-code style.
Other Everyday Applications of AI for Small Business Owners
AI for Lead Generation
There are some fantastic, full-service platforms that do AI lead generation (like Braze). They’ll do everything from identifying leads that match your target criteria to using predictive analytics to anticipate your customer needs to testing and optimizing ad campaigns.
Another way to go about lead gen is by using AI-powered SEO tools. SEO stands for Search Engine Optimization and essentially stands for the algorithms that rank websites on Google. There are various ways to impact SEO, like the keywords you use on your site, how fast your pages load, and how well-organized your site is. Here are some great SEO tools that leverage AI to help optimize your spot in search engines. These vary a bit in how they work and what they do, but for the most part, they will either audit your existing site and help tell you what you can do better, or they’ll help you write new content so it’s optimized for SEO.
Finally, AI can be used in multiple ways to create great targeted ads. It can help you analyze customer data and behavior patterns to understand the characteristics and interests of your target audience. It can then be used to write content and ads to increase your chances of reaching and connecting with those exact people.
AI for Better Decks
Beautiful.AI is an excellent one-stop shop for the average deck creator. It can get a little pricy, especially for teams, but if you’re replacing Canva, it will fit into your budget seamlessly.
I use a couple of different forms of AI to help create my slide decks.
- I ask ChatGPT to create an outline based on the information I have readily available when I start.
- I create all my notes in Notion, which has an AI feature that helps me flesh things out.
- When I start my slide deck, I use Canva’s AI to create titles, captions, etc.
Check out Botsonic to build your chatbot — no-code style.
How to Get Started with AI
Full transparency here, my biggest struggle with using AI is finding the right tools to do what I need because there are so many available now. My best advice for you is to:
- Ask ChatGPT! Remember that it’s limited, especially by its knowledge timeline, but it’s valuable.
- Utilize Product Hunt — I linked it in the database, but Product Hunt is a fantastic way to find new products of all kinds, and they specifically have an AI section. It is limited to more recent products but still very worthwhile.
- Stay up-to-date on AI by subscribing to newsletters and following relevant topics on LinkedIn, Twitter, and news sites. Even if you skim things, you’ll start picking up on what people use.
- Talk to other business owners and find out what they’re using! We’re all figuring this out together; feel free to reach out to people who might be doing similar things and get an idea of the tools they like.
The Million Dollar Question
Is AI going to take my job?
No — AI will not take your job. But a person who knows how to use AI will.
I am firmly convinced that AI isn’t going to take our jobs or put innovative companies out of business. The people — students, employees, and entrepreneurs alike — who can leverage AI in combination with their own knowledge will use it to get ahead. The people who rely solely on AI to get answers and do their jobs will end up sorely regretting it and potentially losing their careers over it. The companies currently replacing copywriters and other skilled humans with AI will end up crawling back once they discover how low-quality the output is in comparison.
The only way to get good, helpful, content-aware responses from AI is to understand your topic on a fundamental level before going in. Otherwise, it’s just Ask Jeeves.
AI will continue to improve and be able to give better answers. It might replace the jobs of people who do data entry and other menial tasks that don’t require critical thinking in order to be successful, but it’s not a big scary monster that’s going to overtake industries and destroy careers. AI isn’t actually designed to replace us; it’s meant to help us perform tasks and get information more efficiently.
That being said, the people who understand AI and are willing to learn how to use it to their advantage will get ahead and turn into big scary monsters that overtake industries and destroy careers. AI is here, it’s powerful, it’s useful, and it’s not going away, so we have to get on board.
My Favorite AI Tools as a Small Business Owner and Entrepreneur
- Calendly for scheduling
- Otter.ai for note-taking and transcription
- Grammarly for grammar and spelling in all my documents
- Chatbots for customer service
- Monday.com for project management
- Notion — my entire life lives on Notion, if you haven’t tried it I can’t recommend it highly enough, and they have recently upped their AI capabilities
This article is directly based on a presentation I gave to the Huntington, WV Chamber of Commerce on May 31, 2023.
Ariana Shives is a social entrepreneur and product designer passionate about design thinking and social impact. Her work is focused on education, fintech, and startups and small businesses. Ariana is the Co-Founder and Head of Product Design at Kairos, a fintech startup on a mission to provide simple, relatable, proven financial education and resources to those underserved by traditional financial resources.
Ariana graduated from the University of Southern California in 2017 with a Bachelor’s degree in Neuroscience and in 2019 with a Master’s degree in Social Entrepreneurship. She currently serves as an Entrepreneur in Residence at Marshall University’s Center for Innovation and Entrepreneurship (iCenter) and as the West Virginia Advocate for Right to Start.
She can be reached via email at: [email protected]