AI Chatbots Aren’t Taking Your Jobs: They’re Your New Assistants

A Pragmatic Journey into GenAI as Your Personal Assistant

I have never been one to jump on most technology bandwagons early; I am very pragmatic about what technology can do rather than what it promises. This extends to generative AI. I was not the first to play with ChatGPT and Gemini when they came out in the early 2020s. Maybe it’s because I work in fields that use machine learning very effectively. Even though I was aware of the leap Google made in 2012, I wasn’t eager to dive into the new wave of AI when it first appeared.

Now we’re in 2025, and the models are improved, the applications expanded, and integrations increased. Perhaps most significantly, the progress seems to be plateauing, as reflected by the Gartner Hype Cycle for Generative AI.

After reviewing AI summaries in recorded meetings, reading anecdotes, and attending a few webinars, I thought maybe it’s time I investigate GenAI chatbots. What finally pushed me to explore GenAI was watching young hackers at a capture-the-flag competition. They jammed every challenge into a chatbot for quick fixes—an impressive, if slightly chaotic, display of AI-assisted problem-solving. While I am never the first to jump on, I also don’t want to miss the bandwagon altogether.

So here I am four months later with Gemini Pro added to my Google One subscription, a Microsoft Copilot license through work, and a ChatGPT account.

Your Job Isn’t Going Anywhere — Yet

First things first: in my humble opinion, unless your job is to answer the same question a thousand times, your job will not be replaced by an AI chatbot. There are a vast number of jobs out there that still need very human qualities like empathy, critical thinking, and creativity. AI attempts to generate creative content such as images and videos still seem to lack, and usually require modification and polishing.

Will your job be replaced in the future? Maybe. Only time will tell. There is still a lot of growth in how AI is applied across the board, and vendors are still working on new and improved models.

What about “now” now? The best way I have seen GenAI used so far is as a supplement to the existing workforce. Companies are leveraging GenAI for tasks like summarizing text and generating code—accelerating workflows without replacing the human touch. Much like the printing press allowed a single skilled person to crank out leaflets for the masses faster than an army of monks with quills, ink, and parchment, generative AI enables us to create content summaries and code suggestions faster and cheaper than asking another person for help. Public statistics show that unemployment is low, but job growth is slowing down. Social media also reflects the same, with an abundance of laid-off workers and recent college grads taking more than six months to find a job. AI is supplementing current employees, reducing the need to hire more people.

Actual people are still necessary to correct inconsistencies and fix mistakes, and we still need the humanity and color that make our content worth consuming. All these points point to the truth that, even though your job isn’t getting eliminated in favor of GenAI, you should investigate the AI tools available.

Why You Should Learn to Use GenAI

You’ve made it this far, and want to use GenAI. That’s great! But what should you use it for?? As I said earlier, GenAI is great for summarizing or suggesting updates to existing knowledge. I will spare you the gritty details, but basically, GenAI Chatbots uses a Large Language Models to ingest prompts and other content and transform it into a response. A large language model (LLM) is a statistical algorithm that returns the most likely string of words based on the input string of words.

Interacting with a GenAI chatbot is an interesting exercise. Think of talking to a GenAI chatbot like talking to an autistic child (I say this as the father of one). You must be very specific in what you ask for. Give lots of details and avoid assumptions, colloquialisms, sarcasm, abbreviations, and any other references that require cultural knowledge. Set clear expectations up front. The GenAI response is only as good as the input provided. Just like vague questions return vague or wrong answers, clear and specific prompts return better and more specific responses. Also, don’t forget that GenAI chatbots have no emotions or humor. While they are designed to replicate human speech patterns, those patterns are simulated, not genuinely felt. GenAI chatbots do not understand your feelings or the feelings of others whose content they’ve been trained on. GenAI chatbots simply recognize patterns in prompts and that those patterns resulted in other patterns from other people. What is commonly referred to as a chatbot hallucination just means that the chatbot took an unexpected path through the model. The more specific you are with what you ask, the less likely GenAI is to take the wrong turn at Albuquerque.

Finally, just like someone you just met guessing wrong that you’re a New England Patriots fan (go Baltimore!), the style of your chatbot’s response is only as good as the references it has to what you like. Use the thumbs up to let it know you like how it responded or give it a reference to your preferences in past writings or a very direct statement.

What Can AI Chatbots Actually Do?

Once you get used to using AI chatbots, you’ll be surprised by what you can get from them! Just like all new endeavors, it takes practice. If you think of the common jobs that high-powered executives ask a personal assistant to do, you get a list like:

• Schedule appointments
• Take dictation
• Draft emails, presentations, documents
• Research
• Coordinate meetings
• Schedule travel

With access to the Internet and integrations with SaaS services, chatbots can do all of these tasks with varying degrees of success. If you were to hire a human to perform these tasks every day for you at $15 an hour, you would pay $2400 a month. The cost of an entry-level license of a GenAI chatbot costs $10-$20. Every company charges more for extra features, but no matter what, there’s a price range well within your reach. As the title says, GenAI chatbots are the personal assistants you’ve never been able to afford!

Let’s jump into some examples!

Example 1: Therapy Support for My Son

This first example was a personal project, so I used Gemini and a Google product called NotebookLM.

My autistic son goes to a weekly therapy session where he gets social group therapy. During this therapy, the therapists spend time collaboratively with four other boys, working through exercises to understand each other’s reactions to events by observing demeanor and listening to changing speech patterns. Frequently, the therapists use gameplay, like Lego building and board games to facilitate interactions. Recently I suggested using tabletop role-playing games (TTRPG) like Dungeons and Dragons as a therapy tool for older kids and teens. I recently found (through AI searches) a paper published on the topic, as well as a TTRPG designed to integrate with therapy. While reading the paper I quickly realized I didn’t understand most of the psychological references, and this paper was 60 pages chock-full of the stuff. I decided to test AI.

NotebookLM is a tool that integrates the Gemini chatbot with a notebook like Evernote or Obsidian. It enables you to upload or link references that focus on what the chatbot references to answer your questions. I asked the chatbot to summarize the paper for someone who doesn’t have any psychological experience. Based on the chatbot’s response, I asked more detailed questions about individual topics. After I had one of my son’s therapists validate the summary and I validated the individual topic explanations myself with regular searches, I felt confident in its response. I was able to summarize the TTRPS’s gameplay to help the therapists out. I was even able to take it one step further: I found another TTRPG that was simplified for a younger audience. I added the rulebook as a reference and had the chatbot summarize how to apply therapeutic principles and some guided gameplay sessions for the therapists!

Example 2: Writing This Article

The second example affects the article you are reading right now! Since this is more work-related, I decided to use Copilot.

I am a bit of a thought leader at my company, so I write blog articles and make social media posts for my online followers. I have seen several of my colleagues show summaries or papers that AI generated, which they then modify. That seemed neat, so I tried my hand at it. To be honest, I didn’t like what was generated. For the most part it was factually correct, but it didn’t sound like my voice. This is when I found that my Copilot license enabled me to create an agent to assist me on specific topics and activities. Microsoft offers an starter template for a writing coach. I uploaded copies of every piece I previously wrote and added this configuration prompt:

Once saved, I was ready to start interacting. I had a marketing specialist generate a blog post about my primary product with a generic AI agent. Again, I didn’t like how it was written and decided to use my newly created Technical Blog Ghostwriter (TBG) to make improvements. I copied and pasted the entire content into my TBG as a prompt, then asked it for suggestions. After a few iterations, the content got to an acceptable place; it captured my voice enough to publish.

I found this process a bit difficult to go through, as I had what AI wrote stuck in my head when trying to rewrite it in my own voice. I find editing someone else’s writing difficult apart from using AI. When someone tries to write something on my behalf, I have a hard time getting past the existing writing to make it my own. So it took several iterations to get it right. Since then, I found that I prefer doing my own writing and have my Technical Blog Ghostwriter work as more of a copy editor, suggesting changes that make sense for my writing style. That is exactly how this article was written! I wrote the entire first draft without help, but I pasted the content into my trusty TBG to help clean it up and add some flair that is in character. In any case, it was the iterations and determining what I liked and didn’t like that helped set the tone for my Technical Blog Ghostwriter. Just like it takes weeks for a high-paced executive to train a new personal assistant, so does the chatbot.

The Downsides of AI Chatbots

I would be remiss if I didn’t talk about the negative aspects of the AI Chatbots. First, the energy consumption required to run these large language model-based chatbots is massive. This isn’t surprising, considering we’re trying to simulate a thinking brain instead of just programmatically performing pre-set tasks. Companies have worked hard over the last few years to optimize power consumption, but this jump still has real-world consequences. Many of the companies building and running AI chatbots in the cloud are rushing to build their own power plants for fear that a jump in use could cause a catastrophic power shortages (think not enough power for hospital ventilators or elevators). This increase in power consumption must come from some source, and clean energy may not be enough to supplement, so companies are investing in nuclear and fossil fuel power. A friend asked me if GenAI is worth boiling the world over for? That’s a good question.

The next issue I see is the potential for younger generations to become dependent. The same anecdote earlier about the young hackers jamming every problem into the Chatbot for a quick solution felt lazy to me. They ran as fast as possible because, well, it was a competition, but when asked if they understood what was being exploited in the challenge, they gave me a blank stare. Just like we had to learn to differentiate between reliable information about our 16th president being a secret vampire hunter, we need to learn when and when not to trust AI and use its references as a source of what to learn next. If you are familiar with the movie Idiocracy, you know what I am talking about.

Final Thoughts

While it is highly unlikely your job will be eliminated in favor of a GenAI chatbot, there is no telling what will happen in the future. Now is the right time to start your foray into using AI. There are so many training courses available for every AI company and chatbot in existence. Many of these courses were created by individuals just like me who jumped in and played around. Companies like Udemy, Coursera, and Google all offer courses, as do many universities like MIT, Johns Hopkins, and Stanford (many of them are available for free!). If you’ve been curious but hesitant, let this article be your invitation to explore GenAI. The tools are ready—and so are you. Enjoy the ride!