Ten Steps for Tapping Into the Potential of AI

One of the most noticeable technology trends shaping our world today is undoubtedly the rise of artificial intelligence (AI). Both traditional AI and generative AI promise to make their mark on our industry, and our world. In fact, at the rate it’s advancing, AI could possibly be the single most impactful technology evolution in our lifetimes. 

As someone who was a pioneer at the forefront of the shift to the cloud, I have lived through massive tech transformations and learned that change waits for no one, and this is no exception. If you aren’t already looking at how AI can help you gain efficiencies in your business, you’re already behind. Let’s explore the different types of AI along with 10 recommendations for how to begin incorporating AI into your business strategy. 

Traditional AI analyzes sets of data and makes informed decisions or predictions based on those inputs. A fitting example of leveraging AI for better decision-making comes from the restaurant industry where a well-known chain triangulated weather data, local sports team schedules, and historical sales data to predict future sales. Upon calculating the predictive AI results, the restaurant was able to shift inventory to specific sites on sunny game days and maximize profits in significant ways. 

Traditional AI is already in use by companies in our industry in value-generating areas, such as support. For example, we map historical data sets across hundreds of thousands of devices to make highly informed predictions about device health to stave off support calls even before they occur. This provides a better customer experience for clients, while at the same time reducing costs.

Separately, generative AI (GenAI) draws from massive amounts of human language-based datasets (large language models) and creates completely new content or visual outputs. This form of AI is the most widely talked about today because it’s advancing very quickly, mimics the outputs of humans, and has incredibly significant legal and even moral implications. I’ve attended over a dozen of the top technology conferences throughout the past year and GenAI was the focus of them all. 

What’s both exciting and frightening is the speed of GenAI advancement as well as the potential impact. When people ask me why I believe these advancements are happening so rapidly and the rate of investment is so high, I tend to compare the technology shifts of the past to a combustion engine, while GenAI is like an electric one. Take cloud for example. While it has undoubtedly reshaped how we deliver IT, it took a bit of time to make that transition. First, cloud providers needed hardware/software/services to be in place in data centers worldwide and then they had to get businesses to displace the on-premises solution they already owned. Therefore, the speed of transformation was not overnight. Like in a combustion engine, you need air, gas, compression, and spark before you can create enough energy to propel the vehicle. It’s rapid, but not instantaneous. 

Yet with GenAI, the barrier to entry is almost zero — you may need no specialized infrastructure (other than an internet connection) and no existing expertise since it’s a conversational, human language-based interaction. Therefore, like an electric engine, you can propel the vehicle almost on command. With almost ubiquitous accessibility and applicability, it’s understandable that we’re talking about exponential growth. And the financial potential, both in terms of recent technologies as well as creating efficiencies within existing environments, is also just as impressive. According to the 2023 McKinsey & Company report titled “The Economic Potential of Generative AI,” the productivity advancements of generative AI could add trillions in value to the global economy, potentially more than the entire GDP of the United Kingdom! 

It’s not all upside, however. GenAI’s human-like interaction is also where stark moral and legal challenges reside. If a machine can mimic a human, even in image and audio, how do we know what’s real and credible? A notable example of this was a research project I undertook using ChatGPT — I asked it to provide the top five facts on a specific topic and the facts it returned were exactly what I needed to complete my research. However, I then asked it to provide detailed sources for each fact and it cited a mix of real publications that I could verify, as well as a few fake sources with publication dates in the future. Clearly, all the data was not credible, yet it sounded completely real. 

Furthermore, if anyone can take advantage of such powerful technology without any training, how do we ensure that what they create is used for good and not evil? I’m not even talking about dark web levels of evil, although that is certainly a real threat, but something less nefarious and business-like. For example, I was on an AI webcast a few months ago and the speaker asked for product ideas in the chat tool, which he then put into an AI tool that auto-generated a full business plan for that product. About six months later, I saw that same product concept advertised for sale. Although impossible to tell, it did make me wonder, did we put someone’s idea at risk by putting it into an open AI tool so that someone else could steal that idea and turn it into profit? 

Lastly, how will we be able to identify true expertise if it’s all coming from AI? I can assure you that it would have been 100x faster to have asked ChatGPT to write this article for me, but then I would have lost all credibility as a thought leader because the thoughts I took credit for were not my own. And, as a trusted resource for my clients and partners, don’t I owe it to them to be the one doing the thinking? The person face-to-face with you in a meeting should be the same as the person behind the screen authoring this article. But how will we know who to trust if not everyone takes that same approach? 

These are just a few small examples of why thoughtful leadership is a requirement for bringing AI into your business. Given the fact that you can start using AI today with little to no barriers to entry, where do you start? Here are some recommendations based on analyst best practices and my own experience leading an AI Task Force at work that will help you approach your AI journey. 

1. Put guardrails in place. Start by setting strong guidelines for your organization’s use of AI due to security, intellectual property, and copyright risks. Here are a few suggestions for policies to consider putting in place: 

a. Proprietary intellectual property and personal information should not go into any online AI tools. If your proprietary data would benefit from AI, then consider setting up a private instance. 

b. Do not download or use software in conjunction with internal company software until receiving written approval from your IT department.

c. Do not claim content generated by AI as your own. 

d. Management should expressly approve any AI tool. 

2. Set up an internal cross-functional AI team. Bringing in leaders from across the organization will help avoid data silos or investment duplication and will perpetuate best practices more quickly. Plus, it helps to ensure strong oversight to enforce the guardrails you established in Step 1. 

3. Gather potential use cases. Closely collaborate with team members to discuss which tasks they may be able to leverage AI to complete. Have them consider their current day-to-day repetitive tasks, as well as novel areas that could add value. Most importantly, emphasize that the organization stands by team members and helps them function more efficiently while expanding their roles in other areas. Act transparently in this way. AI could be seen as a threat to employee’s jobs, so make sure it’s seen as a value-add to enable higher-value work. 

4. Identify use cases that will provide high impact. Rank those use cases based on high/medium/low impact on the business. This will help you narrow the list of projects to focus resources on ideas that might provide the highest value. 

5. Assess effort. Upon identifying high-impact use cases, estimate the effort it would take to implement as high/medium/low priorities. This will help you further narrow the list for quick wins versus longer-term investments.

6. Identify toolsets. Across your top impact use cases, research which tools you should use and purchase if necessary. We found that many groups were already looking into the same tools but didn’t realize it. As soon as we uncovered the common tools, we were quickly able to reduce spending while improving results. 

7. Understand your datasets. Bad data in will result in bad data out. Vet and aggregate your datasets and identify sensitive data. Use cases involving sensitive data may need to utilize private tools or they may need private data fields to be removed and replaced. 

8. Partner. Many of the companies you do business with today, whether it’s your back-end email tool or the company whose hardware and software you distribute, are likely exploring AI. Ask about their plans and roadmaps. It may be faster to extend the tools you use today with AI capabilities than to start from scratch. 

9. Implement and iterate. Implementing and testing AI can be very rapid, but it’s not unlike hiring new employees. You’ll still need to train, coach, and inspect the results. Evaluate small subsets and quickly iterate on the results. 

10. Validate. No matter which use cases you choose, always validate the results. As we discussed before, the human-like outputs can be very reassuring, but also misleading.  

No matter how you implement AI for your business or clients, it promises to bring a host of new opportunities to the world. By taking these steps, you’ll be able to explore the vast potential of AI and embark on a journey of success — systematically and safely. If you haven’t already started exploring how AI can boost efficiencies and create new models, now is the time. 

Toshiba | Posts

Kerstin Woods is Vice President, Solutions and Outbound Marketing, Toshiba.