Why_AI_Free_Deliverables_Don_t_Exist

The transcription discusses the challenges professionals face with the increasing presence of artificial intelligence (AI) in the workplace. It highlights how AI is already integrated into various tools and platforms we use daily, making it impossible to completely opt out of AI in our work processes. The focus is on the myth of being able to create AI-free deliverables and the ethical dilemmas it poses. The speaker, Andrew Jacobs, emphasizes the importance of human judgment and accountability in work outputs, despite the pervasive use of AI in the background. Jacobs suggests a defensible approach to using AI tools for source discovery, where AI generates potential sources, but humans critically evaluate and select relevant and reliable information. The discussion delves into the blurred line between human-created and AI-generated work, emphasizing the need for professionals to navigate this gray area ethically. So think about your own week for a second. Just put yourself right in the middle of your standard workflow. Oh, I'm sure we've all been there recently, right? So maybe you're submitting this massive strategic project for your boss, or I don't know, maybe you're on the other side of the table and you're hiring a freelancer to write some marketing copy. You might just be signing a standard vendor contract and right there buried in the expectations or like highlighted in bold in the fine print is this strict non-negotiable mandate. And it just says no AI, which is everywhere right now. I mean, organizations are just frantically drawing this line in the sand. Yeah. Presented as this firm ethical boundary line. Exactly. And the implication is really simple. If you cross that line, you're cheating, but if you stay on the safe side, you know, you're doing authentic, valuable human work, but the reality of trying to actually stay on that safe side is incredibly complicated. Oh, absolutely. And that brings us to the core of today's deep dive. We are exploring this brilliant excerpt and it's titled the myth of the AI free deliverable. A great piece. It really is. It was written by Andrew Jacobs. He's a strategy and capability development expert, and he also hosts the women talking about learning deep dive. What I love about Jacobs is he brings this highly practical, just completely unsentimental perspective to the conversation. He really doesn't pull any punches. No, not at all. He's not interested in the philosophy of like what AI might do in 10 years. He's looking at how professionals actually operate at their desks on a daily basis right now. Right. The reality on the ground. So our mission today is to completely deconstruct that myth. The idea that you can simply opt out of artificial intelligence at work. Because you can't. You really can't. We want to help you navigate that incredibly blurry, often really frustrating line between ethical tool use and outright cheating. Because, okay, let's unpack this. Let's do it. Picture this scenario. You're a chef, right? You walk into your modern stainless steel commercial kitchen, grab your absolute favorite carbon steel knife, and you spend the next two hours meticulously hand chopping vegetables for a complex mirepoix. Okay. I'm picturing it. So a customer walks into the back, sees the knife and says, wow, I love that you don't use electricity to cook. Right. And you just nod, accepting the compliment, completely ignoring the fact that the massive refrigerator behind you is humming the way the industrial ventilation system is cycling the air. Yeah. The high lumen lights overhead are perfectly illuminating your cutting board. Exactly. Claiming, I don't use AI in the modern workplace is exactly like that chef, claiming they don't use electricity just because they chop their carrots by hand. That is such a good analogy. And what's fascinating here is how Jacobs forces us to confront the literal technical reality of our digital environment. He really drags us out of the theoretical. Yes. To bust the myth of the AI free existence, he moves us away from this idea of opting in. I think the common perception is that AI is this specific destination, you know. Like it's a website you have to go to. Right. Like you have to consciously navigate to a specific URL or explicitly launch an application when you decide you want to automate a report. But Jacobs points out that this is fundamentally a misunderstanding of how the technology is integrated. Because the environment itself is already compromised. Yeah. Or well, enhanced, depending on how you look at it. Exactly. He lists it out so clearly. AI is already operating inside your search engine. Yeah. It's baked into the word processor you use every day. It lives in your internet browser, your grammar checker, your research platform. It's literally everywhere. He has this fantastic core point in the text where he says, you didn't choose to work in an AI enabled environment, opening your laptop, put you in it. Boom. That completely shifts the conversation from a conscious moral choice to just an inescapable environmental reality. And if the operating system itself is using machine learning algorithms to manage your battery life or predict your search queries, you are already participating in an AI workflow before you even open a document. But wait, if artificial intelligence is literally the digital air we are breathing, the second our operating system boots up, how can a client or like a boss legally or even logically mandate no AI? It's a great question. Because if it's baked into the basic spellcheck of the document I'm typing in, aren't they demanding something that is technically impossible? It feels like writing a contract that says, do not breathe oxygen while completing this assignment. Yeah. That technical impossibility, it exposes a massive gap in our professional vocabulary right now. How so? Well, when a client or a manager writes no AI in a contract, they aren't actually talking about your workflow. They aren't saying, make sure you disable your browser's predictive search, or I want you to turn off the squiggly red line under your misspelled word. Right. They don't care about the grammar squiggles. They fundamentally do not care about the minutiae of your workflow. They are talking exclusively about the final deliverable. So they are only looking at the finished product sitting on their desk, completely ignoring the assembly line that built it. Exactly. They're trying to draw a boundary around the genesis of the core ideas and the final execution of the work. They just want assurance that a human brain generated the core value. But because our tools are so saturated with these background algorithms, the definitions of what constitutes work are shifting rapidly, aren't they? Oh, at lightning speed. Jacobs points out that the phrase human reviewed is currently being thrown around as this sort of professional reassurance. Oh, I see that everywhere now. It has become a corporate buzzword meant to calm clients down. You hand in a report and you attach the phrase human reviewed to signal safety. But Jacobs explicitly notes that human reviewed is absolutely not the same thing as human created. Not even close. Here's where it gets really interesting. Because I look at that phrase human reviewed, and I immediately want to push back on it from a client's perspective. Yeah, it's a very loaded phrase. Isn't human reviewed just a very polite, sanitized way of saying the AI did all of the actual heavy lifting. It generated all the ideas, it wrote all the paragraphs, but hey, I gave it a quick thumbs up at the end before I hit send. Pretty much. Yeah. It feels like a loophole big enough to drive a semi truck through. If I'm paying premium agency rates, I don't just want someone skimming a machine's output. And the loophole is massive. That is exactly why most modern professional practice currently sits in this very murky, really uncomfortable gray area. We are trapped somewhere between entirely human created and merely human reviewed. Which is a tough spot to be in when you're trying to prove your worth. Definitely. Clients and commissioners of work are ultimately trying to buy human accountability. They are paying for your unique authenticity, your lived experience, your specific expertise. Right. They don't necessarily care if you use a 1990s style analog workflow or sitting in a library with physical highlighters and printed encyclopedias, but they are absolutely terrified of paying thousands of dollars for something a generative model spit out in four seconds. They want to know they are paying for your judgment, not just a processing fee for server space. Exactly. Which brings up the crucial dilemma Jacobs tries to solve. If the laptop itself is inherently AI enabled and the client fundamentally wants your human judgment, how do we actually operate in that gray area? Like what are the defensible ethical ways to use this environment? Yes. And Jacobs focuses heavily on one specific area that he considers completely defensible and that is source discovery. Gathering the raw materials before you start building? Yes. To understand why this is defensible, he compares the old way of doing things to the new way. For years, knowledge workers relied on RSS aggregators. Oh wow, throwback. Right. For those who might not remember, these were tools like Google Reader in the late 2000s or early 2010s. Instead of manually visiting 50 different blogs or news sites every morning, you would subscribe to their RSS feeds. And a bag reader would just push all those articles into one single dashboard. It gathered the information for you. Precisely. Jacobs argues that using an AI tool like Perplexity today is just a significantly more informed advanced version of that old aggregator. Okay. So if I'm using AI to build my reading list, how does Jacobs define the workflow to keep it ethically sound? Because, I mean, an AI search engine operates very differently than an old RSS feed that just gave me everything chronologically. The distinction really lies in who evaluates the information. Jacobs outlines a very specific defensible process. Okay. Walk me through it. In this workflow, an AI tool like Perplexity identifies candidate documents. It scours the web, understands the semantic context of your query, and it returns a lead list of potential sources. A lead list. Okay. But here is the critical boundary. The human must then manually open every single source on that list. Wow. Every single one. Every single one. The human checks the primary document, reads it, and uses their own expert judgment to decide whether it is actually relevant, reliable, and useful for the project. As Jacobs puts it, the AI returns the leads, but the human does the work. I have a really practical question about that though. Go for it. If you still have to manually open, read, verify, and cross-reference every single candidate document on that AI-generated lead list, does an AI search tool actually save you any time? That's the million-dollar question. Right. Or does it just give you a much wider, much more overwhelming starting line? I feel like I'd spend hours just vetting the AI's suggestions, dealing with hallucinations, or just totally irrelevant pangents, whereas a targeted Google search might've just given me the three PDFs I actually needed. If we connect this to the bigger picture, the goal of this defensible workflow isn't necessarily about saving raw time in the short term. Really? Then what is it about? It is about shifting cognitive load and elevating the quality of the final output. The AI acts as a tireless high-speed assistant for the heavy lifting of gathering. Okay. I can see that. It can pull semantic connections across diverse disciplines that you might never have thought to search manually. The critical factor is that the human retains 100% of the cognitive load of evaluating. Right. Evaluating is the human part. You might spend the exact same amount of total time on the project, but you are spending it analyzing a vastly superior, more comprehensive set of data, rather than burning your mental energy just trying to find the data in the first place. You're outsourcing the legwork so you can preserve your energy for the brain work. Beautifully said. You let the machine run into the library and grab 50 books off the shelves, but you are still the one sitting down to read the chapters and synthesize the arguments. That is the core of defensible use. The judgment remains fiercely human. Okay. But once we establish that gathering sources is acceptable, Jacobs forces us to define exactly what crosses the line into the unethical. Where must the human take over completely and what actions are strictly indefensible? Yeah, this is the part everyone struggles with. Knowing where it's okay to let the machine help and where it becomes academic or professional dishonesty. Where does Jacobs draw the actual red line? He draws it very clearly around the manipulation of the final output. Jacobs lists the specific indefensible actions. It is indefensible to use AI to draft, rewrite, edit, refine, or proofread, and then not disclose it to the client. Wait, so what does this all mean? Because I am spotting a massive contradiction here based on what we established just a few minutes ago. Earlier, Jakin said that grammar checkers are just part of the inescapable, acceptable environment of our laptops. They're the background noise. Right. But now he is saying that using AI to proofread without disclosure is explicitly indefensible. Where on earth is the line between an automatic spell checker fixing my comma splice and an AI proofreader refining my sentence? This raises an important question, and it really gets to the heart of what constitutes intellectual value. To understand the difference, we have to look at the mechanics of how these tools actually work. Okay, break it down. The line between a grammar checker and an AI proofreader is the line between structural mechanics and stylistic intent. Meaning one follows a rule book and the other makes choices. Essentially, yes. A traditional grammar checker is deterministic. It is a rules-based system operating off a static dictionary and established syntactic rules. Like catching a missing comma. Exactly. It flags a misspelled word or a missing comma. It corrects the raw mechanics of the language without altering your intellectual framework. It really doesn't care about your tone. But generative AI is different. Very different. Generative AI operates on probabilistic models. It doesn't just check rules. It predicts the next word based on vast data sets. Right. So when you ask an AI to refine or proofread a paragraph today, it is invariably restructuring your argument. It is choosing different verbs to alter the tone. It is smoothing out your unique voice and replacing it with the statistical average of its training data. It's making editorial choices, not just spelling corrections. It is actively participating in the persuasion of the text. Yes. The words are merely the vehicle for your ideas. Your judgment, how you choose to phrase something, what you emphasize, the specific tone you strike, the rhythm of the sentences that is the actual product being purchased. So if a probabilistic AI is making those editorial judgments and you don't tell the client, you are essentially misrepresenting the origin of the expertise. Precisely. So you are handing a client a meal and claiming you spent hours developing the complex flavor profile when really you just bought a pre-made sauce from the store and microwaved it. Exactly. The client might think it tastes fine, but they are paying you for your culinary expertise, not your ability to operate a microwave. The deception lies in the origin of the value. And because that line between a basic rules-based grammar checker and an advanced probabilistic AI editor can be so fuzzy in modern software, Jacobs argues there is only one way to resolve this tension moving forward. What's his solution? The only solution is his ultimate standard, which is radical disclosure. Which means the old defense of simply waving your hands and saying, I didn't use AI is completely dead. It is obsolete. Jacobs states plainly that saying I didn't use AI simply won't cut it anymore because as we established in the beginning with the chef analogy, you did use it. Right. It's in your browser, your search, your tools. Denying it just makes you look oblivious to your own environment. So the new standard is full disclosure. This means agreeing on specific boundaries with your client before the work even begins. It means keeping detailed logs of your workflow. And most importantly, it means ensuring what Jacobs calls auditable use. Auditable use. I mean, that sounds like a bureaucratic nightmare if I'm being honest. It sounds heavy. Yeah. What does that actually look like for you and me when we are turning in a project on a tight deadline? What metrics do buyers actually need to see to consider it auditable? It requires a high level of specificity. A client or a manager reviewing your work needs to see, okay, which exact tool was used? Was it chat GPT, Claude perplexity? I need the name. When was it used during the process? Was it during the initial ideation phase or the final polishing phase? What specific task was it used for? And finally, how is the output integrated into the final work? That is intense. I can't just put a little asterisk at the bottom of the page that says large language models were utilized for research purposes. Not even close. Jacobs highlights that this is not just about citing research the way you would cite a textbook in a bibliography. Radical disclosure requires providing the specific queries you ran. It requires logging the exact prompts you used. Wow. Furthermore, you have to document the instructions and parameters you put in place to prevent the AI tool from overreaching. Okay. I have an analogy for this. And it is giving me serious flashbacks. Oh boy. Let's hear it. This new standard is exactly like being in high school algebra and the teacher writing, show your math in giant red letters at the top of the test. Oh, absolutely. Yeah. You can't just write down X equals 42 and expect to get an a, even if 42 is the objectively right answer. The teacher doesn't just want the final deliverable. They want to see the sequence of steps you took to arrive there. Right. They need to see the formula you applied. Exactly. To prove that you actually understand the underlying concepts and crucially to prove you didn't just copy the answer from the kids sitting next to you. The stakes here are obviously so much higher than a high school math test, but the principle is identical. If you look at the macro level themes, Jacobs is tagging in his work, things like AI ethics and AI governance. This radical disclosure is the foundation of the future economy of knowledge work. Buyers fundamentally need to understand how and where AI was used to justify paying for the work. Boy, I have to push back on the reality of this. Sure. In math, there is generally one structured way to get to 42. In writing copy or like drafting a business strategy, there are a million ways to prompt an AI. It's messy. It's iterative. Very iterative. You try a prompt to give you garbage. You tweak it. You try again. If I have to log every single prompt and parameter, that introduces a massive amount of friction into my creative process. How does a client even evaluate a messy prompt log? Are they going to sit there and read three pages of my chat history just to accept a two page report? It's less about them reading every single word of your chat history and more about the transparency of the methodology. The friction you are describing is real, but Jacobs would argue it is necessary friction. Necessary because it proves the value. Exactly. If a client is paying a premium for human insight, the math you show in your prompt logs is the proof of that insight. It proves you directed the tool with expert intent rather than blindly accepting whatever the tool spit out first. Your prompt log demonstrates your strategic thinking. It proves you were the head chef orchestrating the kitchen, deciding exactly how the ingredients should be prepped rather than just hitting the start button on a microwave. This is what Jacobs believes we absolutely must be discussing with commissioners, contributors, and vendors right now. The era of silent utilization is over. Yeah. You have to stop pretending the tools don't exist. And start meticulously documenting how we wield them. It demands a much higher level of professional maturity. This has been such a journey today, taking us all the way from the complete myth of an AI free existence, breaking down the inescapable reality of the algorithms running in the background of our laptops and landing on this new, highly documented reality of the auditable AI workflow. It's a lot to take in. It completely shifts how I think about my daily tasks and specifically how I talk to clients about my process. It demands a much higher level of self-awareness about our own processes. We can't just run on autopilot anymore. So to you listening right now, whether you are managing a team or grinding out deliverables as a freelancer, the main takeaway from Jacobs is incredibly clear. Stop pretending AI isn't in your workflow. It is. The environment demands it. But remember your actual value to your boss, your clients, or your partners lies in your judgment, your curation, and your specific human intent. It does not lie in your typing speed or your ability to generate generic text. Your expertise is the filter through which the AI's output must pass. If you remove the filter, you remove your value. Think about your own workplace tomorrow morning. How does your boss or your biggest client actually define a deliverable right now? Are they still operating under the illusion of no AI? Are they still asking for hand-chopped carrots while standing under fluorescent Exactly. And more importantly, look at your own habits. Are you quietly hiding your AI use because you feel a vague sense of guilt about it? Or are you going to pioneer this new habit of radically documenting it? It is a critical transition. We are moving from hiding a perceived roughness, using a tool to save time, to showcasing a modern strength, which is mastering the tool to elevate quality. And I want to leave you with one final provocative thought to mull over. Something that builds on Jacob's idea of this auditable workflow, but pushes it a little further into the future. Oh, I'm curious. If radical disclosure and meticulous prompt logging become the absolute gold standard for all professional work. If we are all showing our math by handing over logs of our exact parameters, strategic guardrails, and iterative queries, will those prompts and parameters eventually become more valuable to a client than the final written deliverable itself? Are we entering an era where knowledge workers stop selling the meal entirely and start making their money by selling the recipe? It is something to seriously think about the next time you open your laptop and start a new project. Thank you so much for joining us on this deep dive. We will catch you next time.