Using Gen AI to Accelerate Business

Announcer:
Welcome to TriNet's National Small Business Week Summit, where you will hear from thought leaders and experts throughout this five-day event, discussing topics from AI and compliance to culture and HR. Before we begin today's sessions, we have a few housekeeping items to help get you familiar with the virtual event space.

If you look to the right of your screen, you'll see our Q&A tab, where you can put your questions. We have a polling tab where you can participate in polls. And we have a resources tab where you can download relevant information for this session. Unfortunately, our presenters cannot share their slide decks.

At the bottom of the screen, you'll see emoji reactions, options for closed captions and an HD tab if you're having connectivity issues and you wish to downgrade your stream quality. If you do wish to visit one of our information booths during the session, you can click on schedule at the upper right-hand corner and choose the specific booth you would like to visit.

These presentations are for educational purposes only. With decades of experience supporting small and medium-size businesses, TriNet has a unique insight into HR best practices for businesses. TriNet does not provide legal, tax, or accounting advice. Reproduction in part or in whole is not permitted without express written authorization from TriNet.

In our rapidly changing business environment and evolving digital landscape, there are few conversations that do not involve artificial intelligence, AI. From automating routine tasks to delivering personalized customer experiences, gen AI has the potential to transform various aspects of your company.

Here to explain how, we welcome Jeff Hayward, TriNet's Chief Technology Officer, to our virtual stage.

Jeff Hayward:
So thank you for that introduction. And hello, everyone. I'm excited to be talking to you today about one of my favorite topics, technology, and how we can use it to solve business problems and accelerate our business.

As part of our Small Business Week discussions, we thought it was the perfect time to discuss the topic of generative AI and how it is being used and considered by SMBs today. I'm personally excited about the implications of gen AI for small business as it has a great potential to give SMBs the resources to compete with their much larger corporate counterparts.

Today we'll discuss the excitement, the confidence and how SMBs are using this technology to win more customers and compete in more ways. But at the same time in our surveys of SMBs, we're also seeing that there is some fear, uncertainty and doubt in their minds as well. So today we're going to be talking about how small businesses can use generative AI to accelerate business.

In a rapidly changing business environment and involving digital landscape, there are few conversations that do not involve artificial intelligence, AI, the emergence of generative AI, which is the upcoming wave of AI-powered solutions designed to enhance business processes can present never before seen opportunities and concerns for companies seeking to gain a competitive edge. So if we're going to talk about generative AI, let's be sure to define it. Just in case you're not familiar with it. There are many productized versions of the latest generative AI capabilities out there. You may have heard of ChatGPT or Google Gemini, for example.

These are both generative pre-trained transformers, which is a type of large language model that is trained on a massive data set of text and software code. How they basically work is a three-part process. First, they are trained on a massive data set. GPTs are trained on an enormous amount of text data, like books, articles and software code.

This exposes them to various writing styles and linguistic patterns. Second, is that they start to understand the language relationships. Through this training, the model learns to recognize relationships between words and how they are used in a sequence. This allows them to predict the next word in a sentence, the next sentence in a paragraph and so on.

It sounds simple, the ability to predict the next word or sentence, but it really is incredibly powerful. Then in the third phase, they develop the ability to generate new text. Once trained, GPTs can use their understanding of language patterns to generate new text or software code. They do this by considering the prompt or the question you provide as a starting point and then they predict the next word, the next sentence, the next paragraph that follows to create a coherent response to your question.

You know that autocomplete on your phone? That one that guesses the rest of your sentence or word? It's like that. It's the most advanced autocomplete that exists, as it generates entire sentences and paragraphs based on the context of your prompt. The inner workings of these complex models are unclear, however, making it challenging to understand how they arrive at their output.

This lack of transparency can compromise trust and raise concerns among their users. Fostering transparency and explainability in AI development is crucial for its responsive use. So why is this the case? First is scale and complexity. These models involve billions of parameters and intricate connections that are difficult to untangle and analyze.

It's like trying to understand how the human brain works by examining every individual neuron. The next challenge is the training process itself. The training involves massive amounts of text, data and the model learns by identifying patterns and statistical relationships within that data. Exactly how the model uses these patterns to perform tasks is not always perfectly understood.

And finally, you have the black box problem. This lack of transparency into the internal workings makes it challenging to explain how specific output is generated. It's like observing a black box that produces a certain output for a given input, but the internal workings are hidden from us. But despite these challenges, researchers are actively working on methods to make large language models more interpretable.

This is an ongoing area of research and advancements are being made in this area. So in a recent Gartner survey, 21% of CEOs stated that AI is the top disruptive technology likely to significantly impact their industries over the next three years and that a fear of missing out may drive behaviors.

While this is true for large corporations, we wanted to also understand how SMBs with comparatively smaller teams, time and budgets are thinking about this technology evolution, so we conducted our own survey on The Wing, which is TriNet's exclusive community where HR leaders and TriNet admins can meaningfully engage with TriNet to network, collaborate, learn and receive rewards and recognition while also having some fun.

We posed several questions to our SMB community in the survey and we discovered that while 40% of our 60 respondents have not yet implemented AI in their day-to-day business, 32% of respondents have successfully implemented it in their operations today, and another 28% are considering AI technologies, although they have not yet implemented it.

This compares nicely to a recent survey of SMBs that GoDaddy did, where nearly three-quarters of small business owners responded that they've tried generative AI tools, with 26% of them saying that they already are using it in their work environment. So given the tremendous opportunity, we wanted to understand this better.

And when we dug a little deeper and asked what use cases they're considering, we saw that overwhelmingly the response was to use AI to free up staff for other high-value activities. And we definitely agree with this. And we see many potential competitive advantages for SMBs with this technology. A few examples could include streamlining operations by automating repetitive tasks.

In this way, gen AI can automate various tasks, reduce the need for manual labor, facilitate innovation and potentially lower operational costs for SMBs. For example, one way gen AI can boost operational efficiency is by taking advantage of a virtual assistant to not only monitor customer communications, but be able to generate summaries of conversations and help produce contextually appropriate content for follow ups.

And there are many use cases applicable to SMBs, such as AI-powered chatbots that can handle routine inquiries, provide product recommendations and even assist in completing a purchase on your website. This can improve customer satisfaction and reduce agent costs.

It can also be a great real time coach that can help improve your customer service interactions, especially as it has the capability of doing sentiment analysis in real time to aid your agents and not only understanding the optimal interaction, but also to understand the context and adjust as necessary as the call progresses. And as a technologist, of course, I'm certainly excited about the advantages that gen AI brings to software development where developers can see up to a 40% gain in productivity when working on new greenfield software development efforts and opportunities to achieve 10 to 15% in productivity advancement for maintaining and expanding existing systems.

It's also a great technology for improving code documentation and automating the writing of tests. So when we asked about the use cases that our SMBs are using and considering, the overwhelming responses were around content creation, and specifically for content creation for customer service and for marketing and personalization.

This really isn't surprising. Because marketing automation has reached new heights with the integration of large language models, enabling the creation of highly personalized content that's tailored to individual customer preferences or behaviors. And this approach not only increases prospect and customer engagement, but it's also optimizing marketing strategies in real time.

And these powerful AI-based tools are really redefining what's possible in the realm of digital transformation by generating human like text, images and even software code, providing businesses with the ability to automate complex tasks, create rich, engaging content and really deliver exceptional customer service.

But while the opportunities are tremendous, there are also several challenges to consider. Potential problems to consider include the following: data can sometimes be of poor quality, inaccurate or out of date. AI systems can be expensive, complicated or time consuming to implement. Costs can be high if businesses lack in-house skills and need to outsource.

And there remains ethical concerns over racial, gender and other biases in AI data, as well as legal concerns over the collection of sensitive data by AI algorithms. And our survey responses also reflected this, with the two biggest concerns being lack of technical expertise and data limitations. And I believe the respondents are absolutely correct to have these specific concerns.

Fortunately, the hyperscalers like AWS, Microsoft and Google are aiding all of us on the data front by doing a large portion of the training for us and then giving us the ability to move their models into a secure space in the cloud where we can both test the application of these models, plus augment them with our own much more modest data. It's also been important in software to have a solid buy versus build strategy and the use of large language models is really no different. And fortunately, there's a growing marketplace for already trained baseline domain specific models. Growing in-house technical expertise and experience or leveraging it from a consulting or technology research firm is really equally important, as these tools can have equally beneficial or detrimental effects for their users.

So, while it is true that generative AI can boost the performance of highly skilled employees by as much as 40%, according to new studies by professors from schools like Harvard Business School, the Wharton School, the Warwick Business School, MIT Sloan, these improvements make gen AI the shiny new tool for business leaders to consider. However, understanding the right ways and areas to apply them as well as helping your teams feel comfortable using them, it's quite a different story. In further studies, it was found that those using AI technology for a task that is well suited for it performed 40% better than those not using a tool.

However, those using GPT capabilities for a task outside of its capabilities actually performed 13% worse than the control group. Interestingly, when the participants received some training on how GPT tools work, they performed a little bit better when using it for the right tasks. However, even after training, they performed 24% worse when using it in an area it wasn't designed for, again, versus the control group. So although tools like chat GPT can astound you with their ability to craft content across so many industries and domains, it is similar to other software technologies we've deployed over the last several decades.

It still remains important to map the right business processes to the right tool capabilities. What should I do now? How do I use generative AI to accelerate business? The first recommendation is to not fall victim to the hype or to the FUD, the fear, uncertainty and doubt. Hopefully you are convinced now that gen AI is something important to incorporate into strategy, but there's also a real risk to proceeding without understanding.

For the FUD, you can reassure your teams that this is not a strategy to replace them. In my 30-year career, technology was going to replace me as a software developer many times. In the 80s, it was COBOL, the Common Business Oriented Language. In the 90s, it was case tools. In the 2000s, it was no code, low code environments.

And not only did any of these tools fail to replace me and my team, they actually helped us grow our business and teams and they made us more productive to able to get more products out to the market and take on more customers. The other part of this recommendation is to not get caught up in the hype cycle.

These technologies are on the lower end of the maturity curve and there are many challenges. Like the ones I mentioned earlier, which are further compounded by the risk that the AI can hallucinate and give unpredictable and incorrect answers. You may have seen in February, Google had to pause its Gemini image generation of people as it was creating not only inaccurate, but sometimes offensive images.

And we saw this in our risk, we saw the same risk in our survey, as well as the majority of respondents said that misinformation and error rates were really their biggest risk. So let's not rush to implement new vendor tools before we understand them. There's a real risk to brand and other more serious issues if you don't thoroughly understand the tools and their risks.

It's also important to consider the limitations of AI tools, not only in terms of accuracy, but also bias, privacy and security. And remember what I mentioned earlier in the statistics. While various gen AI tools can really help improve productivity 14 to 40%, depending on the domain and use case, it can also contribute to a productivity loss of 13 to 24% if you select the wrong tools for the wrong job.

You wouldn't use a shovel to pound a nail. And similarly, each of these new AI capabilities coming to market have their own strengths and weaknesses and need to be evaluated like you would any other software tool. The other recommendation is to look at AI in this phase of evolution as an advisor to your experts.

If you look at the statistics from MIT, Harvard and other similar institutions, you see that they are most valuable in the hands of your existing experts and that they can make them significantly more productive, enabling you to either do more with less or scale your business larger or faster.

From there, it's necessary to start to bridge the knowledge gap. I've found in this latest wave of gen AI, that technology, research and consulting firms are really especially helpful right now, as they see a very wide range of customers and use cases, and they can help you sort through the hype cycle of what's real and what's still emerging.

Some of these research firms regularly publish public blogs or host free public information sessions. If these are insufficient and the core services a little out of the financial budget, I've learned a lot from local and online technology events that you can see listed on sites like Meetup.com. A quick search of AI events today found over 50 of them occurring over the next few weeks online, even in my local area.

These are great ways to learn how others local to you or online are actually using these capabilities in their business today. Thank you. And as we found that 32% of our survey respondents are already implementing AI technologies, if you are a current or future TriNet customer, we would welcome you to join The Wing community and share ideas on how AI could or is benefiting your business with fellow community members.

So I wanted to thank you for joining me today to talk about the many exciting ways that gen AI is helping SMBs accelerate their business and I wish you an exciting Small Business Week. If there are questions that we didn't answer or you would like to speak to someone, you can go to our information booth.

Thank you.