Artificial intelligence is the broad term for technology that is capable of doing everything from finishing your sentences to writing a novel. The simplest example is the ability of your smart phone to suggest the next word in your text message.
The application in your phone has created a database of sorts of the words you commonly use (or that commonly follow other words). That's how it knows to suggest "Berkeley" if you often send a message to your friend, suggesting lunch in Berkeley; and it might also suggest "me up" after you write "pick." The app doesn't use magic or thought—it uses data.
We use many other more sophisticated types of AI. You can pay your bills by phone using prompts from an automated voice, because the application uses a series of pre-programmed questions and responses, such as "Do you want to pay with the credit card on file?" to guide you through the process.
Siri is another instance of AI; when you ask for the name of the best Italian restaurant in town, Siri canvasses online reviews and distills them into a few sentences. The pop-up messages on websites, offering to help you, are nothing other than a response to what you have so far on the site (or have not done) . The app figures out what you need by placing you among the thousands of users who have already visited the site (and ended up buying something!).
As you can see, AI isn't always the stuff of science fiction movies. It's often as simple as a collection of information that the application consults when asked a question.
Yet newer, more sophisticated applications, like ChatGPT, can write a screenplay, design a clothing line, or respond in a way that's sometimes indistinguishable from talking to a human. These newer applications, known as generative AI, can learn and improve as new data is added; but the more they learn to do, the more questions are raised about what they should be doing.
The use of sophisticated apps raises questions as to the app's accuracy and, in some contexts, the ethics of using it. The sections below explore the challenges of using apps for various tasks in a small business.
AI applications fall into one of three categories, from the simplest to the more complex:
Let's look more deeply at each of these increasingly sophisticated uses.
Small businesses have been using AI to book appointments, order items from a menu, take payments, and answer customer questions like, "When will my order ship?" These uses are fairly noncontroversial, with the possible exception of the ubiquitous "How can I help you?" pop-up. Many people find these pop-ups intrusive and they don't trust them to really know what the customer wants or needs.
AI can help with both simple and complex marketing tasks such as:
Scheduling and implementing follow-up strategies, like drip campaigns. Let's say a customer visits your website and requests an e-book, or other information you offer. You can use an AI application to automatically send additional emails to the customer tailored to their area of interest.
Targeting specific groups of customers. AI can identify and group customer segments, to allow you to customize messages for each segment based on their preferences and needs.
Writing blog or social media posts, landing pages, and emails. AI can automate messaging such as emails for prospecting new business and sort and direct responses to the appropriate person or department for follow up.
If you don't have the time or staff to write your own website blogs or social media posts, AI can write them for you. Some applications are even able to generate topic ideas.
Analyzing advertising responses and customer surveys. If you've ever used customer surveys to uncover problems with your products or services, you know that getting responses is only half the battle. The other half is analyzing the responses and identify ways you can improve. AI can analyze and categorize responses, to help understand what's going right or wrong with your systems or products.
When you advertise on social media like Facebook, you are already receiving AI analytics. They tell you how many people clicked on your ad and give you other information to help you evaluate the effectiveness of your advertising.
A number of AI applications are designed to let sales people focus on selling instead of paperwork. They can accomplish tasks such as:
Prospecting for new business and generating leads. AI can reduce or eliminate the time it takes to chase down leads by sending outreach emails and categorizing and prioritizing responses for the sales team.
AI applications allow your sales team to focus on buyers most likely to make a purchase in the near future, and to deploy strategies to follow up with longer term prospects.
Qualifying buyers who respond to outreach campaigns. AI can ask potential buyers simple questions, like whether they are looking to make a purchase in the near term or how much money they expect to spend on a purchase. The answers to these questions help the sales team to prioritize the calls they make and reduce the chances that they'll spin their wheels needlessly on buyers who are not the right fit for your products.
In general, AI applications that handle back-office and administrative tasks are designed to eliminate the need for manual data entry. They can:
Monitor and manage inventory. Automating inventory management gives businesses real-time information on merchandise that needs to be replenished and reorders that should be processed, so they don't miss out on sales opportunities.
Store and sort customer information. Applications can route customer information to the proper files and update databases, by making changes to customer addresses and the like.
Process orders and invoices. AI can automate your business's processes, from requisition to invoicing, and integrate these procedures with accounting and inventory management software.
Process and issue refunds. Using AI to process and issue refunds eliminates the need for a person to monitor and respond to phone inquiries, and usually shortens the lead time for processing refunds.
Review documents for missing or incomplete information. If your business relies on numerous forms that you and your customers must review and sign, AI can replace the need to manually review the forms by alerting you when a document is incomplete.
Before you begin interviewing vendors who supply AI software, it's a good idea to learn the terms the technology industry uses. Once you've clearly defined what you want the technology to do, these terms will help to refine your online search for the best suppliers for your needs.
While AI can, in some areas, replace human workers, this technology is most often used to free up time for workers to focus on more essential aspects of their jobs.
For example, salespeople can use AI to manage their prospect database and update it for each customer contact, freeing up time to contact the most likely buyer prospects and close sales.
Some additional reasons to use AI include:
Provide faster customer service. Automating processes like placing orders and handling returns means customers don't have to spend time waiting on hold for a live representative.
Reach more customers. Suppose you want to notify your customers of a new product you're introducing. Automating your email system allows you to send more emails than a human worker can in the same amount of time.
Avoid mistakes, oversights, and miscalculations. Using AI will eliminate the mistakes that typically occur when you are manually entering information for databases and recordkeeping. Automated systems can accurately track inventory and reorder promptly to avoid running out of stock.
Improve productivity. Using AI for marketing analytics helps to understand where your advertising dollars are producing the best returns.
Perform tasks you don't have the resources to accomplish. AI can help small businesses make use of social media when they have neither the time nor human resources to manage these types of campaigns.
No matter how much you spend on AI, it's likely not going to be a standalone solution for your business's every need. To succeed, even the best AI solutions will require human intervention at some point in the process.
For example, AI that generates leads for your business won't do a lick of good if the sales team doesn't make follow up calls to the most likely buyers.
AI that analyzes the results of your marketing reports won't help you decide where to best put your advertising dollars if no one reads the reports the automated system provides.
AI that is premised in an outdated data scoop will not give you accurate results.
Likewise, when AI isn't able to handle more complex customer inquiries or complaints. You won't generate much customer goodwill if you don't include an option to talk with a live agent.
When you are purchasing AI tools, you must first decide on the tasks you want the software to accomplish. Each application is designed for a specific purpose, and, In some cases, for a specific industry. For example, automated ordering software designed for restaurants is different than software designed for ordering wholesale merchandise.
The companies that offer software you already use are a good place to start when you want to add AI capabilities. These same companies might offer AI tools you can add on, or they might partner with other companies that offer AI software. When you start your research with vendors you already use, you'll also be certain that the new tools you purchase are compatible with your existing software.
Your research should also explore the degree of technological expertise required to use the application and the training and support the vendor offers.
Like many IT solutions, AI is sold in off-the-shelf applications for hundreds of dollars or less, or you can customize solutions for hundreds of thousands of dollars.
The more complex the chore that you want AI to complete, the bigger the tech stack (series of technologies stacked one over the other) required to build the app. Accordingly, those apps are more expensive than a simple tool that auto fills a repeat customer's orders.
Email management software is the least expensive of AI applications. Vendors typically offer the software for a monthly subscription at costs that range from $10 a month to a few hundred dollars a month.
The cost of chatbots varies widely, depending on the capabilities you want.
When you are budgeting for the cost of implementing AI, remember to include the technical support and training you'll need in addition to the software licensing fees.
It's also a good idea to pay attention to the integrations a vendor offers (the ability of one company's software to interact with software from other companies). Choosing a vendor that offers numerous integrations allows you to expand your AI capabilities by adding new software down the road when your needs change or expand.
Following common sense rules, such as involving human oversight into your use of AI, can help avoid common problems associated with AI. As noted earlier, it's important to give customers the option of talking to a real human, so they don't become frustrated when an automated response doesn't solve their problems.
However, the most serious problems with AI, and the reason it has become so controversial, is that the technology is quickly advancing to a point where it will be difficult, if not impossible, to distinguish between images and information that are computer generated and those created by humans.
It won't be long before AI technology is able to generate a completely lifelike avatar that can be used to sit in for the human it was designed to replicate on a video conference. Imagine the consequences if one of these avatars is substituted for the CEO of a powerful organization and provides shareholders, the press, or others with false information.
Left unchecked, AI can be used by bad actors to steal identities, bully the unsuspecting, spread misinformation and fake news, or release propaganda for political or competitive purposes.
Even when no harm is intended by the user, AI can make mistakes. A mayor in Australia recently threatened a defamation lawsuit if the developers of a popular AI application didn't quickly fix an erroneous report that named him as a participant in a bribery scandal. He was, in fact, the whistleblower who reported the criminal activity to authorities.
One of the most widely recognized flaws of AI is its tendency to fabricate information when it doesn't have all the data it needs. The result can be libelous as exemplified above, misleading, or just plain weird.
In one instance, an AI program used by the Los Angeles Times became confused and reported that a large earthquake had just occurred in Santa Barbara. It turned out that the earthquake in question took place in 1925.