By Chris Tamm, Managing Partner of Cast Services, Inc., an AI and Automation implementation firm.
Artificial Intelligence is all the rage in business talks, influencer marketing, and company press releases. Having completed many cross-industry AI and Automation implementations, I have seen first-hand how the technological trajectory in all industries, including payroll and benefits administration, is changing with increased adoption and implementation of AI and Automation.
Artificial Intelligence (AI) stands at the helm of this transformation, offering a plethora of tools and capabilities that are profoundly impacting how professionals navigate the complexities of their domains. By embracing AI you stand to significantly reduce operational costs and also unlock unprecedented scalability to meet the dynamic needs of a growing business landscape.
The automation of routine tasks, predictive analytics for better financial planning, real-time compliance monitoring, and fraud detection are just the tip of the iceberg when it comes to cost savings. On the other hand, data-driven decision-making, streamlined onboarding processes, enhanced employee experiences, and capacity-building herald a new era of scalability that is vital for maintaining a strong competitive position.
The early adopters of this technology are setting a precedent that will likely redefine the benchmarks of operational efficiency and growth in this sector, making AI an indispensable asset for modern-day payroll and benefits administration professionals. This window of opportunity changes every day, and once everyone has implemented the technologies the competitive advantages can be less significant. So it’s important to understand how quickly your organization must embrace or at least be familiar with AI.
Where to Start with AI in Your Organization
When a technology or efficiency can touch as many areas of expertise as AI the opportunities and risks can be overwhelming. When considering implementing pieces of AI into your operations to start gaining the efficiencies, you need to start building the organizational muscle for adapting to an increasing rate of change. You also need to ensure the right conversations are happening, so you stay in control of the narrative, your data, and your competitive advantages.
When you have these in place you can begin implementing a few practical ways to incorporate AI into what you do in these four areas:
1. Security and Data Privacy
Two issues to consider with data privacy involve using someone’s likeness (face, voice, etc.) and how to protect your company’s data. At the time of this writing some of these services use your submitted information for “training” to help their systems get smarter, have more information to work with, understand our behaviors and requests, etc. For most of my requests like marketing content, most email rephrasing, etc., this is fine.
Some systems have ways to prevent this sharing of information. In ChatGPT you can turn off Data Sharing in your Settings, which also removes some features from your experience. Another example is that OpenAI (the company that provides ChatGPT) has an electronic connection (API) to their program which, by default, does not use any information for its training. API is easy to connect to and can be used to have more “private” interactions with your company and personal data with many different types of controls not available in ChatGPT.
Current off-the-shelf tools make it easy to impersonate someone in order to use their likeness (face, voice, etc.). With just three seconds of someone’s voice AI can re-create it. One of the largest increases in corporate fraud uses voice-verified wire transactions. If someone can hear your voice, they can use it. It is also fairly easy to take a picture of someone and animate it into a video or teach an AI to duplicate their face and voice and have it talk as them.
Helping your team understand what is possible will help you defend against any mis-uses. Teach your organization and team members that they can’t use anyone else’s face or voice without their specific permission. The next time your bank calls to verify something, hang up and call them back to make sure it is really them. Use a passcode or passphrase with your family so you know it is really them!
2. How to Protect Your Company’s Data
I recommend adopting an ‘AI and Automation Acceptable Uses’ policy so your team-members can understand your stance on how AI can be used and when it needs to be disclosed. I recommend addressing the following in your policy:
- No use of another person’s face or voice without their specific consent (for specific use cases).
- No use of public tools for confidential information (to understand how data can be used for training with specific programs).
- No use of AI-messaging without ‘Human In the Middle’ for verification (if this is your stance on the subject).
- Ensuring that any AI use does not violate other rules and regulations; for example, using ChatGPT or another AI tool to analyze resumes for a hiring and selection process could possibly discriminate if they ignored certain protected classes or individuals.
- Citations or specific facts cited by an AI tool must be verified before being used in a corporate setting unless the tool has been tested and validated previously by your organization.
- Help your team understand who to reach out to if they want to use AI or Automation and don’t know if their process would be compliant. For example, if you need to work with private data (Personally Identifiable Information) in an Excel sheet there are ways to have AI tools write code to manipulate the Excel sheet without the actual data being submitted to a tool like ChatGPT. Asking is better than breaking your rules!
3. Entry Tools to Learn and Use
Do you and your team know how to use some initial AI tools like ChatGPT? These ‘generative-AI’ tools can help you complete tasks and think about your business in different ways. Do you and your team understand what ‘Prompt Engineering’ is and how it can pay dividends when starting to adopt AI?
To get started, I recommend working through a few different small projects with ChatGPT to experiment with how natural language programming can help accomplish some previously time-consuming tasks. Some things can be accomplished in many ways, and although it can be dizzying to learn them it will help you understand how different approaches can produce very different results.
There are many different services similar to ChatGPT, such as Google’s “Bard” and Microsoft’s “Bing.” In the interest of brevity, I will focus on ChatGPT. For example:
- Go to ChatGPT, sign in, and type in the following: “I need to create a new website for a financial services business, auto dealership, or whatever your business is. Can you please help me create content for the homepage with sections, a blurb of the content for each section, and a description of any pictures that should be included?”
- Once this completes, type (in the same chat) something like, “Thank you. Can you please take #2 and #3 and write the full content including header, call to action, etc?”
- If you have ChatGPT PLUS (paid subscription) start a new chat using GPT4 and select Advanced Data Analysis (turn on via Settings Beta Features) and upload an Excel file. Then type, “Can you please explain the Excel information to me?” You can also ask things like, “Can you please analyze the Excel sheet and forecast the next six months for me and give me back an Excel sheet?”
- Start a new chat using GPT4. Click the picture button in the chat-bar, upload a picture, and type, “Can you please explain this image to me?”
- You can also start a new Advanced Data Analysis task and ask it to “create a PowerPoint presentation for me.” Ask what you want on the first, second and succeeding slides.
4. Prompt Engineering
This relatively new term deals with how to talk to AI-enabled systems that can understand natural language programing instructions. There are many ways to ask an AI system to accomplish a task, and different approaches can have very different results. Most of these systems are like a golden retriever in that they gladly create what you asked for, sometimes with fabrications and hallucinations mixed in! There are ways and systems to safeguard against these un-truths, but those involve a more advanced view than we will cover here.
Hopefully this blog post has provided greater context and specific ways about how to look at AI and Automation for your organization and team(s). There are many other considerations when implementing projects, including ideating alongside your team members, discussing the appropriate guardrails, and then working through the change management of new and evolving technologies. Building the right capabilities and developing an organization that can successfully implement and integrate AI into more efficient and dynamic operations require thoughtful planning, an open mindset, and the right roadmap to execute effectively.
About the Author
Chris Tamm is the Managing Partner of Cast Services, Inc – an AI and Automation implementation firm that helps organizations adapt and implement across many different verticals including government, financially regulated industries, manufacturing, real estate, insurance, healthcare, cybersecurity, legal, education, and more. If you would like to learn more about Cast Services, Chris Tamm, or receive their newsletter please visit www.cast.services or reach out to Chris directly at firstname.lastname@example.org if you need help to successfully implement AI and Automation throughout your organization and across your teams.
The views expressed in this article are those of the author and do not necessarily represent the views of PenChecks Trust®, its subsidiaries or affiliates.