AI’ing AI: Making Generative AI Simple and Accessible

Ed note: Posts published on From the Workplace are written by outside contributors and do not necessarily reflect the view or opinion of SHRM.

As the pace of business accelerates, every organizational function — from HR and marketing to operations and communications — can benefit dramatically from the immediate productivity leap offered by Generative AI (GenAI). 

For example, integrating GenAI into HR functions can boost productivity by up to 30% in the near future, according to a 2023 Boston Consulting Group study. Additionally, one early AI adopter highlighted in their study successfully reduced its annual HR budget by 10% year over year for three consecutive years.

Barriers to Achieving These Benefits: Why Has GenAI Adoption Been Slow?

Despite its considerable promise, the widespread adoption of GenAI has been slower than expected.

One of the key barriers to broader adoption is the demanding technical skill known as "prompt engineering.” Prompt engineering requires crafting very specific, detailed instructions to guide AI tools effectively — a skill typically requiring significant training.

For example, drafting a clear prompt to analyze employee survey data requires knowing exactly how to instruct the AI on scope, analysis method, and desired output format — details that can easily overwhelm users without technical training.

Now, let’s explore some new and exciting ways to significantly accelerate (or even eliminate!) the prompt engineering learning curve.

AI Development for Everyone: Simple Tools That Eliminate Complexity

There is a pathway to simplified GenAI solutions that empower individuals without technical backgrounds.

To understand this, consider the most elite spreadsheet users. These are not the people who teach Excel training courses; rather, they are the end users who have mastered advanced techniques through trial and error in their day-to-day spreadsheet usage.

Our current era heralds the rise of a new kind of ‘AI programmer’ — super users of GenAI who don’t need to master traditional programming languages. Instead, their fluency in GenAI allows them to create sophisticated tools entirely within the GenAI environment, without any traditional coding skills.

In contrast, an AI super-user can quickly create and orchestrate multiple versatile AI tools in parallel, delivering cohesive, customized results quickly and seamlessly — without relying on numerous external vendors or fragmented integrations.

For example, the innovative prototypes featured below—AutoPromptBot and AutoTaskBot—were developed entirely using GenAI platforms, without the need for coding, APIs, or external integrations. This capability effectively democratizes tool and application development, making powerful AI-driven solutions accessible to anyone, regardless of technical background.

Practical Solutions: Two New Prototypes for Simplifying AI

Now let's dive deeper into practical examples highlighting two simple prototypes developed to overcome GenAI complexities.

Here’s how people without programming backgrounds can create powerful tools entirely within the GenAI environment. I call them “autobots,” and they can reduce (and sometimes eliminate entirely) GenAI interactions.

Prototype 1: AutoPromptBot - Simplifying User Interaction

AutoPromptBot is designed for users who handle various distinct tasks within a single domain, such as HR. It does not entirely eliminate interaction with GenAI but greatly simplifies it to the extent that the end user does not need to know how to give instructions (i.e., engineer prompts) to the GenAI platform. Instead, users simply:

• State (or choose from a drop-down menu) their objective (e.g., "I need a performance improvement memo for my employee").
• The tool then provides three simple multiple-choice questions, for which the user selects either A, B, or C, as shown in the table below.
• AI generates a professional response, customized to the user’s answer choices.

For example: "I need a performance improvement memo for my employee."

Result: Immediate, professional output without requiring end-user prompt engineering skills.

This tool is particularly helpful when the end user does not have any other reference documents to attach to the GenAI platform.

Limitations: A key limitation of AutoPromptBot is that its accuracy and effectiveness rely heavily on the subject-matter expertise of the GenAI expert who initially configures it. To ensure high-quality outputs, this individual must have sufficient domain expertise — in this example, deep HR knowledge — to identify the most relevant questions and appropriate multiple-choice responses in the pre-trained model.

Prototype 2: AutoTaskBot - Automating Execution of Predetermined Tasks

This tool eliminates the need for prompt engineering entirely for singular, clearly defined tasks. It’s for people who don’t want to interact with the AI at all, but it can only execute one predetermined task (e.g., “create a job posting from a job description”).

Here’s how it works:

Let’s say a recruiter wants to use AI to create a job posting but is not comfortable with GenAI platforms and lacks prompt engineering skills.

A GenAI expert would us AI to create a text “script file” (no programming languages required) that instructs the GenAI platform exactly how to accomplish this task.

All the recruiter needs to do is:

1. Attach this script file to their GenAI platform.
2. Attach the job description file they want to associate with the posting.
3. Type a few simple words into the GenAI dialog box, such as “follow the instructions provided in these files.”
4. That’s it! The GenAI reads the pre-packaged instructions in the script file and generates a polished, customized job posting.

Limitations: One practical limitation of AutoTaskBot is the amount of data it can effectively handle. While highly efficient for clearly defined tasks, the tool currently works best with concise scripts and instruction files. Extremely large or complex data files may exceed the tool’s current capacity and require additional refinements and testing to determine optimal usage limits.

Natural Extensions of This Work 

• Scaling this process: The GenAI super-user may be able to load enough of these script files into the AI platform to train it to automatically generate scripts for many other topics. 
• Integration: The AutoTaskBot can also easily integrate with AutoPromptBot, providing a delivery mechanism for the functionality that greatly reduces the need for prompt engineering.  
• New, more powerful autobots: While AutoTaskBot is good at simple, single-step tasks, it can potentially evolve to manage more sophisticated, multiphase workflows. By merging AutoPromptBot’s guided interactions with AutoTaskBot’s straightforward automation, users could complete a short form detailing their key project objectives. This form, along with relevant reference files, would be processed by GenAI to deliver comprehensive, high-quality outputs—retaining ease of use while significantly expanding the Autobot’s capabilities. 
• The autobot ecosystem: Welcome to the future, featuring a modular ecosystem of interoperable tools including many different autobots that can be rapidly assembled (think Lego blocks!) and orchestrated, enabling seamless, scalable automation across diverse tasks and domains.

Real-World Examples: Immediate Value in HR 

So, now let’s illustrate the impact of some of these tools with real life applications, specifically within HR:  

• Internal HR program communications: Rapidly produces clear, consistent, engaging messaging. 
• Compensation benchmarking: Reduces manual effort in summarizing competitive pay data. 
• Enhanced hiring and retention: Efficiently analyzes recruitment data for improved hiring decisions. 

Further enhancements could address risk by incorporating built-in data security features such as anonymization scripts, or secure, permission-based access to sensitive HR data. And the underlying design supports scalability; organizations can continuously expand their library of custom prompts and task-specific scripts, facilitating widespread adoption and impact across HR and the organization. 

Simplified AI = Faster Adoption and Productivity 

Finally, removing all this AI complexity: 

• Speeds automation of time-consuming tasks. 
• Upgrades employee capabilities by augmenting value-added tasks.
• Accelerates organizational efficiencies, cost savings, and growth. 

These tools are quite different from ‘vendor brand name’ large scale AI-driven HR platforms which typically require significant investment, complex integrations, and specialized technical resources. Instead, this approach offers agility, cost-effectiveness, accessibility, and immediate deployment, providing AI benefits quickly and easily for almost anyone in HR. 

Ira Feder has held senior human resources leadership positions, primarily focused on sales compensation, for the past 25 years across multiple Fortune 500 companies and industries. Over the last few years, he has focused extensively on creative generative AI tool innovations.