The AI Agents Are Coming! So Are The Reasoning Models. Will They Take Our Jobs And How Should We Prepare?

AI image generated using Google ImageFX from a prompt “Create a digital painting depicting Paul Revere on his midnight ride, but instead of a person riding the horse it is a futuristic robotic AI agent yelling 'The AI Agents are coming for your jobs!'"

Last Fall I traveled to MIT to watch my daughter play in the NCAA volleyball tournament. On the way, we passed signs for Lexington and Concord. AI agents were on my mind. There was a sudden buzz about AI agents and how they’re coming for our jobs. The image of Paul Revere came to my mind.

Instead of warning about the Redcoats stealing munition at Concord, Revere’s of today warn of AI agents stealing our jobs. Then new AI reasoning models released causing another rise in discussion. Like Lexington Green have the first shots been fired on our jobs with reasoning AI agents?

AI image generated using Google ImageFX from a prompt “Create a digital painting depicting Paul Revere on his midnight ride, but instead of a person riding the horse it is a futuristic robotic AI agent yelling 'The AI Agents are coming for your jobs!'
AI image generated using Google ImageFX from the  prompt “Create a painting depicting Paul Revere on his midnight ride, but instead of a person it is a robotic AI agent yelling ‘The AI Agents are coming for your jobs!’.” https://labs.google/fx/tools/image-fx

What is an AI agent?

Search interest in AI agents spiked in January. If you search AI agents Google returns 216 results. Reading through many of them there are probably half as many definitions. For simplicity, I will begin by quoting AI Marketing Institute’s Paul Roetzer, “An AI agent takes action to achieve goals.”

That doesn’t sound scary. What’s driving interest and fear is adding the word autonomous. Roetzer and co-founder Mike Kaput have created a helpful Human-to-Machine Scale that depicts 4 levels of AI autonomous action.

Marketing AI Institute’s Human-to-Machine Scale:

  • Level 0 is all human.
  • Level 1 is mostly human.
  • Level 2 is half and half.
  • Level 3 is mostly machine.
  • Level 4 is all machine or full autonomy.

Full autonomy over complete jobs is certainly fear inducing! Large language model companies like OpenAI, Google, and SAAS companies integrating AI are promising increased autonomous action. Salesforce has even named their AI products Agentforce, which literally sounds like an army coming to take over our jobs! Put some red coats on them and my Paul Revere analogy really comes to life.

Every player in AI is going deep.

In September Google released a white paper “Agents” with little attention. Now, after the release of reasoning models, everyone including Venture Beat is analyzing it. In the paper, Google predicts AI agents will reason, plan, and take action. This includes interacting with external systems, making decisions, and completing tasks – AI agents acting on their own with deeper understanding.

OpenAI claims its new tool Deep Research can complete a detailed research report with references in “tens of minutes.” Something that may take a human many hours. Google’s DeepMind also has Deep Research, Perplexity has launched Deep Research, CoPilot now has Think Deeper, Grok3 has a Deep Search tool, and there’s the new Chinese company DeepSeek. Anthropic now has released what it is calling the first hybrid reasoning model. Claude 3.7 Sonnet can produce near-instant responses or extended step-by-step thinking that is made visible. The Redcoats are coming and they’re all in on deep thinking.

Graphs of Google Trends search data showing an increase in search for AI Agents and Reasoning Models.
Interest in and discussion about AI Agents and AI Reasoning Models has risen sharply. Graphs from https://trends.google.com/trends/

What is a reasoning model?

Google explains Gemini 2.0 Flash Thinking is “our enhanced reasoning model, capable of showing its thoughts to improve performance and explainability.” A definition for reasoning models may be even more difficult and contested than AI agents. This term returns 163 results in a Google search and perhaps just as many definitions.

For my definition of a reasoning model, I turn to Christopher Penn. In his “Introduction to Reasoning AI Models,” Penn explains, “AI – language models in particular – perform better the more they talk … The statistical nature of a language model is that the more talking there is, the more relevant words there are to correctly guess the next word.” Reasoning models slow down LLMs to consider more words through a process.

LLMs and reasoning models are not magic.

Penn further explains that good prompt engineering includes a chain of thought, reflection, and reward functions. Yet most people don’t use them, so reasoning models make the LLM do it automatically. I went back to MIT, not for volleyball, but for further help on this definition. The MIT Technology Review explains that these new models use using chain of thought and reinforcement learning through multiple steps.

An AI prompt framework, such as the one I created, will improve your results without reasoning. You also may not need a reasoning model for many tasks. Reasoning models cost more and use more energy. Experts like Trust Insights recommend slightly different prompting for reason models such as Problem, Relevant information, and Success Measures. Brooke Sellas of B Squared Media shared President of OpenAI Greg Brockman’s reasoning prompt of Goal, Return Format, Warnings, and Context Dump.

Many want a magical AI tool that does everything. In reality, different AI is better for different things. Penn explains generative AI is good with language, but for other tasks, traditional forms of AI like regression, classification, or even non-AI statistical models can be a better solution.

How we talk about AI matters.

Humans are attracted to the magic capabilities of AI. Folk tales like The Sorcerer’s Apprentice which you may know from Disney’s Fantasia, are about objects coming to life to do tasks for us. Reasoning models are said to have agentic behavior – the ability to make independent decisions in pursuit of a goal. Intentional or not, it sounds like angelic, bringing up mystical thoughts of angels and the supernatural.

Since the first post in my AI series, I’ve argued for maintaining human agency and keeping humans in the loop. Therefore, I want to be careful in how I talk about these new “reasoning” models that show us their “thinking.” I agree with Marc Watkin’s recent Substack “AI’s Illusion of Reason,” that the way we talk about these AI models matters.

An AI model that pauses before answering and shows the process it followed doesn’t mean it is thinking. It’s still a mathematical prediction machine. It doesn’t comprehend or understand what it is saying. Referring to ChatGPT or Gemini as it versus he or she (no matter the voice) matters.

Google Gemini 2.0 Flash Thinking
I asked Google’s reasoning model Gemini 2.0 Flash the difference between human thinking and AI “thinking.” From https://aistudio.google.com/

What’s the difference between human and AI thinking?

I asked Google’s reasoning model Gemini 2.0 Flash the difference between human thinking and AI thinking. It said, “AI can perform tasks without truly understanding the underlying concepts or the implications of its actions. It operates based on learned patterns not genuine comprehension.” Does this raise any concerns for you as we move toward fully autonomous AI agents?

Humans need to stay in the loop. Even then, you need a human who truly understands the subject, context, field, and/or discipline. AI presents its answers in a convincing well-written manner – even when it’s wrong. Human expertise and discernment are needed. Power without understanding can lead to Sorcerer’s Apprentice syndrome. A small mistake with an unchecked autonomous agent could escalate quickly.

In a Guardian article, Andrew Rogoyski, a director at the Institute for People-Centred AI warns of people using responses by AI deep research verbatim without performing checks on what was produced. Rogoyski says, “There’s a fundamental problem with knowledge-intensive AIs and that is it’ll take a human many hours and a lot of work to check whether the machine’s analysis is good.”

Let’s make sure 2025 is not like 1984.

I recently got the 75th anniversary edition of George Orwell’s 1984. I hadn’t read it since high school. It was the inspiration behind Apple’s 1984 Super Bowl ad – an example of the right message at the right time. It may be a message we need again.

AI isn’t right all the time and right for everything. It’s confident and convincing even when it’s wrong. No matter how magical AI’s “thinking” seems we must think on our own. As AI agents and reasoning models advance discernment is needed not unthinking acceptance.

The 250th anniversary of Paul Revere’s ride and the “Shot heard ‘round the world” is in April this year. Will AI agents and reasoning models be a revolution in jobs in 2025? In my next post, I take a deep dive into how AI may impact marketing and communications jobs and education. What’s your excitement or fear about AI agents and reasoning models?

This Was Human Created Content!

More Than Prompt Engineers: Careers with AI Require Subject Matter Expertise [Infographic].

This graphic shows that in stages of learning you go through attention, encoding, storage, and retrieval. You need your brain to learn this process not just use AI for the process.

This is the fourth post in a series of five on AI. In my last post, I proposed a framework for AI prompt writing. But before you can follow a prompt framework, you need to know what to ask and how to evaluate its response. This is where subject matter expertise and critical thinking skills come in. A reason we need to keep humans in the loop when working with large language models (LLM) like ChatGPT (Copilot), Gemini, Claude, and Llama.

Photo by Shopify Partners from Burst

Will we all be prompt engineers?

Prompt engineering is promoted as the hot, new high-paying career.” Learning AI prompt techniques is important but doesn’t replace being a subject matter expert. The key to a good prompt is more than format. As I described in my post on AI prompts, you must know how to describe the situation, perspective, audience, and what data to use. The way a marketer or manager will use AI is different than an accountant or engineer.

You also must know enough to judge AI output whether it’s information, analysis, writing, or a visual. If a prompt engineer doesn’t have subject knowledge they won’t know what AI got right, got wrong, and what is too generic. AI is not good at every task producing general and wrong responses with the right ones. With hallucination rates of 15% to 20% for ChatGPT, former marketing manager Maryna Bilan says AI integration is a significant challenge for professionals that risks a company’s reputation.

AI expert Christopher S. Penn says, “Subject matter expertise and human review still matter a great deal. To the untrained eye, … responses might look fine, but for anyone in the field, they would recognize responses as deeply deficient.” Marc Watkins, of the AI Mississippi Institute says AI is best with “trained subject matter experts using a tool to augment their existing skills.” And Marketing AI Institute’s Paul Roetzer says, “AI can’t shortcut becoming an expert at something.”

Prompt engineering skills are not enough.

As a college professor, this means my students still need to do the hard work of learning the subject and discipline on their own. But their social feeds are full of AI influencers promising learning shortcuts and easy A’s without listening to a lecture or writing an essay. Yet skipping the reading, having GPT take lecture notes, answer quiz questions, and write your report is not the way to get knowledge into your memory.

Some argue that ChatGPT is like a calculator. Yes and no. This author explains, “Calculators automate a . . . mundane task for people who understand the principle of how that task works. With Generative AI I don’t need to understand how it works, or even the subject I’m pretending to have studied, to create an impression of knowledge.”

My major assignments are applied business strategies. I tell students if they enter my assignment prompt into ChatGPT and it writes the report for them then they’ve written themselves out of a job. Why would a company hire them when they could enter the prompt themselves? That doesn’t mean AI has no place. I’ve written about outsourcing specific tasks to AI in a professional field, but you can’t outsource the base discipline knowledge learning.

AI can assist learning or get in the way.

I know how to keep humans in the loop in my discipline, but I can’t teach students if they outsource all their learning to AI. Old-fashioned reading, annotating, summarizing, writing, in-person discussion, and testing remain important. Once students get the base knowledge then we can explore ways to utilize generative AI to supplement and shortcut tasks, not skip learning altogether. We learn through memory and scientists have studied how memory works. Used the wrong way AI can skip all stages of learning.

Click the image for a downloadable PDF of this graphic.

I remember what it was like being a student. It’s very tempting to take the second path in the graphic above – the easiest path to an A and a degree. But that can lead to an over-reliance on technology, no real discipline knowledge, and a lack of critical thinking skills. The tool becomes a crutch to something I never learned how to do on my own. My performance is dependent on AI performance and I lack the discernment to know how well it performed.

Research skills in searching databases, evaluating information, citing sources, and avoiding plagiarism are needed to discern AI output. The online LLM Perplexity promised reliable answers with complete sources and citations, but a recent article in WIRED finds the LLM search engine makes things up and Forbes accuses it of plagiarizing its content.

A pitch from OpenAI selling ChatGPT Edu, says, “Undergraduates and MBA students in Professor Ethan Mollick’s courses at Wharton completed their final reflection assignments through discussions with a GPT trained on course materials, reporting that ChatGPT got them to think more deeply about what they’ve learned.”  This only works if the students do the reading and reflection assignments themselves first.

Outsourcing an entire assignment to AI doesn’t work.

A skill I teach is situation analysis. It’s a foundation for any marketing strategy or marketing communications (traditional, digital, or social) plan. Effective marketing recommendations aren’t possible without understanding the business context and objective. The result of that situation analysis is writing a relevant marketing objective.

As a test, I asked ChatGPT (via Copilot) to write a marketing objective for Saucony that follows SMART (Specific, Measurable, Achievable, Relevant, Time-bound) guidelines. It recommended boosting online sales by targeting fitness enthusiasts with social media influencers. I asked again, and it suggested increasing online sales of trail running shoes among outdoor enthusiasts 18-35 using social media and email.

Then I asked it to write 20 more and it did. Options varied: focusing on eco-friendly running shoes for Millennials and Gen Z, increasing customer retention with a loyalty program, expanding into Europe, increasing retail locations, developing a new line of women’s running shoes, and increasing Saucony’s share of voice with a PR campaign highlighting the brand’s unique selling propositions (USP). It didn’t tell me what those USPs were.

Which one is the right answer? The human in the loop would know based on their expertise and knowledge of the specific situation. Generated with AI (Copilot) ∙ July 2, 2024 at 3:30 PM

I asked Copilot which is best. It said, “The best objectives would depend on Saucony’s specific business goals, resources, and market conditions. It’s always important to tailor the objectives to the specific context of the business. As an AI, I don’t have personal opinions. I recommend discussing these objectives with your team to determine which one is most suitable for your current needs.” If students outsource all learning to LLMs how could they have the conversation?

To get a more relevant objective I could upload proprietary data like market reports and client data and then have AI summarize. But uploading Mintel reports into LLMs is illegal and many companies restrict this as well. Even if I work for a company that has built an internal AI system on proprietary data its output can’t be trusted. Ethan Mollick has warned that many companies building talk-to-your document RAG systems with AI need to test the final LLM output as it can produce many errors.

I need to be an expert to test LLM output in open and closed systems. Even then I’m not confident I could come up with truly unique solutions based on human insight If I didn’t engage information on my own. Could I answer client questions in an in-person meeting with a brief review of AI-generated summaries and recommendations?

AI as an assistant to complete assignments can work.

For the situation analysis assignment, I want students to know the business context and form their own opinions. That’s the only way they’ll learn to become subject matter experts. Instead of outsourcing the entire assignment, AI can act as a tutor. Students often struggle with the concept of a SMART marketing objective. I get a lot of wrong formats no matter how I explain it.

I asked GPT if statements were a marketing objective that followed SMART guidelines. I fed it right and wrong statements. It got all correct. It also did an excellent job of explaining why the statement did or did not adhere to SMART guidelines. Penn suggests explain it to me prompts to tell the LLM it is an expert in a specific topic you don’t understand and ask it to explain it to you in terms of something you do understand. This is using AI to help you become an expert versus outsourcing your expertise to AI.

ChatGPT can talk but can it network?

Last spring I attended a professional business event. We have a new American Marketing Association chapter in our area, and they had a mixer. It was a great networking opportunity. Several students from our marketing club were there mingling with the professionals. Afterward, a couple of the professionals told me how impressed they were with our students.

These were seniors and juniors. They had a lot of learning under their belts before ChatGPT came along. I worry about the younger students. If they see AI as a way to outsource the hard work of learning, how would they do? Could they talk extemporaneously at a networking event, interview, or meeting?

Will students learn with the new AI tools that summarize reading, transcribe lectures, answer quiz questions, and write assignments? Or will they learn to be subject matter experts who have discerned via AI Task Frameworks and AI Prompt Frameworks the beneficial uses of AI making them an asset to hire? In my next post, the final in this 5 part AI series, I share a story that inspired this AI research and explore how AI can distract from opportunities for learning and human connection.

This Was Human Created Content!