Do you remember that period when every week someone published yet another ranking of which AI model was better? When LinkedIn was overflowing with posts like “AI will replace developers, lawyers and marketers by the end of the year”? I\u2019m a pragmatist and I wrote posts about it, cautioning people not to believe everything vendors claim. It looks like that\u2019s coming to an end, which is a good thing.<\/p>
What I\u2019m seeing now in the companies I work with at CampusAI is, above all, growing awareness. People have stopped being dazzled by gimmicks and gadgets. Instead, they\u2019ve started asking about things that really matter: how these tools work under the hood, what happens to the data we give them and whether they can be trusted.<\/p>
Importantly, this isn\u2019t a change I see only in corporations or among young AI enthusiasts. I witnessed a conversation at a birding meetup, where a group of retirees were discussing the species of a bird we\u2019d seen in a nearby tree. One woman, around seventy, said she had described it and checked it with ChatGPT. Another, of a similar age, replied immediately: “You have to be careful, because ChatGPT hallucinates and you need to verify that information carefully.” Not bad, right? Seniors who not only use generative AI, but do so thoughtfully and critically. That\u2019s real social change, much deeper than anything I\u2019ve seen at industry conferences.<\/p>
Let’s set aside the grand visions for a moment and talk about something concrete. If someone acquires basic skills in using generative AI tools\u2014even readily available ones like ChatGPT, Google Gemini or Anthropic Claude\u2014they can save at least an hour a day. Sometimes two. On writing emails, drafting documents, structuring thoughts or preparing presentations.<\/p>
What difference does one hour make? Let’s do the math. <\/p>
If you have 10 people on your team and each earns 60 USD per hour, then if each person saves just one hour every day thanks to AI, we’re looking at savings of 600 USD per day\u2026 and that’s 4,800 USD per week. Not much?<\/p>
In that case, if you don’t teach people to use AI effectively at work, you’re losing over 57,000 USD per quarter on manual clicking, writing, research and analysis\u2014tasks that simple AI tools in the hands of trained people can do much more efficiently and often better. I save over an hour a day\u2026<\/p>
Let\u2019s take this article you\u2019re reading right now. It came about in a rather unusual way. I interviewed myself\u2014I recorded my thoughts in the Fireflies transcription tool, fed the text into Claude to help me create an initial structure. Then I sat down and worked on the text myself, refining the details, checking facts and making sure it sounded like me. Once again, to be clear: this is not an article written by AI. One tool turned my stream of words into text. The other organized it, removed repetitions, cleaned it up and created a structure. I wrote the final text.<\/p>
And that\u2019s exactly the model I\u2019m seeing among more and more people around me. AI as a superpower, not a replacement. A tool that helps you get started, create structure, gather information and build a framework\u2014so that people can then take it into their own hands and ensure quality. That\u2019s what we\u2019ve been promoting at CampusAI from the beginning. Artificial intelligence should amplify what we do well, not replace us. When someone steps into an area they don\u2019t understand and lets an algorithm replace them (laziness), nothing good happens.<\/p>
There’s one change that fascinates me in particular, because I’m a software engineer by training. Thanks to vibe coding tools, people who have never written a line of code are now creating working applications. An idea becomes a prototype in a few hours (and sometimes even in 15 minutes). The prototype quickly becomes a tool that the team starts using the next day.<\/p>
I personally know lawyers who have built their own tools for contract analysis. An auditor friend of mine built tools to analyze huge collections of ISO documents and audit reports that he used to spend hours on. I know people in marketing who have built applications that automate their daily work. A few years back, this would have been unthinkable, because it required either a lot of time and people, or a lot of money. Today, above all, you need curiosity and a few hours with the right tool.<\/p>
Agentic systems are evolving at a mind-boggling pace. More and more companies are building solutions that integrate with the applications we use every day. Tools like Anthropic Claude Code and n8n show where all this is heading. 2025 was heralded as the \u201cyear of AI agents,\u201d but it turned out to be a year of preparations for a true revolution that, by all indications, will begin in 2026!<\/p>
But (and this is an important “but”) we\u2019re not yet at the point where you can always take your hands off the wheel. Trust in these systems is growing slowly\u2014and rightly so, because hallucinations, misinterpretations and unpredictable outputs are still commonplace. Much of the automation that used to rely on simple “if X then Y” rules is now successfully handled by agentic systems, but it remains under human supervision in important areas. It will be some time before we can leave them running unattended in critical processes, but the promise is enticing, because the return on investment will be enormous.<\/p>
I often get questions on LinkedIn about which AI tool I use. The answer is simple: lots of them, depending on the task. The features of these tools change from week to week. Every now and then some vendor surprises with something great or, unfortunately (lately even more often), disappoints by degrading a feature I was just using while developing another that doesn\u2019t matter to me, but does for investors.<\/p>
Anthropic Claude is my first and unrivaled choice for writing, because it writes naturally and feels less \u201cplasticky\u201d than the competition. Google Gemini integrates very well with Google Workspace, and I really like their Canvas, which is great for longer-form work. ChatGPT, on the other hand, is really good at research and synthesizing information, and it does so very thoroughly and neatly\u2026 Although you have to know how to ask! You don\u2019t have to pay for all of them at once, since you can switch between subscriptions month to month, though at the expense of your chat history, unfortunately.<\/p>
At the outset, I mentioned that I’m a pragmatist. I’m also a realist. I’m fascinated by technology and innovation, but to be fair, I also have to talk about my concerns. Human nature is driven by laziness, or as someone put it, by “energy-use optimization.” The availability of AI tools makes people take shortcuts. The internet is being flooded with content completely generated by AI\u2014texts, graphics, videos\u2014that look professional but add no value. The phenomenon already has a name: AI slop.<\/p>
Because of this, I\u2019ve become much more wary and I look at every image, text and video very carefully. I verify information more thoroughly than ever, which takes more time and makes everyday media use harder. I quit most social media (good riddance, Instagram!), because I got tired of AI-generated content whose only goal is to evoke awe or outrage\u2014strong emotions designed to keep me on the platform as long as possible.<\/p>
Video generation tools like Google Veo and Kling create incredibly realistic content in a matter of minutes. That’s (frighteningly) wonderful technologically, but the side effect is that many creative people are losing their jobs because consumer expectations have dropped. We’re content with material that’s hard to even call creative, because all it offers is a cheap dopamine hit and time for “not thinking.” Maybe I’m naive, but I believe in a renaissance of true human creation. Good content will stand on its own, because we’ll need the truth. It will prevail, although it will probably cost much more than the AI-generated kind. Yes, you can already see that media outlets want to charge for human-written articles, while giving us AI slop mixed with ads for free. Sad, but that’s the price of progress. This will also get regulated at some point.<\/p>
There’s also the matter of security. AI in the wrong hands is a powerful tool for phishing, information theft and manipulation. This is a real threat that we need to talk about openly, especially in the context of training employees.<\/p>
First of all: cultivate your curiosity. As a tech practitioner with ADHD, I have it easier. I dive into all the new things with curiosity; I explore them, learn and test their capabilities. You should also test different tools, explore and probe their limits. This is really the best way to know what to expect from them. Sometimes the tasks you do every day can truly be done faster and better with AI, but you need an open mind to figure out how to do it. It’s not always obvious. Curiosity!<\/p>
Train your teams. Seriously. I keep seeing that basic generative AI skills are still far too low. Companies are exposing themselves to a loss of credibility and reputation because employees can\u2019t ensure the quality of the materials they produce. Do you remember the case of Deloitte in Australia that sent AI-hallucinated documents to a client? That should not happen, especially at a firm of that caliber. Training was missing, and there was no change to the quality-control system. Change is a process. That process is called \u201cinnovation adoption.\u201d<\/p>
In my opinion, 2026 is the year of a pragmatic approach to generative AI. We are (all of us, we technical folks too) tired of investor-tailored promises and successive versions of tools designed to wow the mainstream media and stir things up on morning TV. No more sensationalism, no more swooning over yet another set of benchmarks. We’ve become more aware, more mindful of security and more demanding of toolmakers.<\/p>
However, I notice that many companies lack structured education and employee engagement in the smart use of artificial intelligence. And that’s a major role for management, which should set an example. Because if the boss doesn’t use AI consciously, every day, and wisely, why would their people do it?<\/p>","protected":false},"excerpt":{"rendered":"
A 70-year-old woman at a birding meetup warns a friend that ChatGPT hallucinates. A lawyer is coding his own AI app. And after three years of working with almost a hundred companies, I can tell you one thing: we’ve finally stopped getting carried away and started working.<\/p>\n","protected":false},"author":252,"featured_media":17666,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_lock_modified_date":false,"footnotes":""},"categories":[799,832],"tags":[],"popular":[],"difficulty-level":[36],"ppma_author":[632],"class_list":["post-17708","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-lifestyle-2","category-editors-picks","difficulty-level-easy"],"acf":[],"authors":[{"term_id":632,"user_id":252,"is_guest":0,"slug":"bartosz-dobrowolski","display_name":"Bartosz Dobrowolski","avatar_url":{"url":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/03\/GRZEDZINSKI_20241022_GRZ_8847-scaled.jpg","url2x":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/03\/GRZEDZINSKI_20241022_GRZ_8847-scaled.jpg"},"first_name":"Bartosz","last_name":"Dobrowolski","user_url":"","job_title":"","description":"Ekspert transformacji cyfrowej i Head of AI Enablement w CampusAI. Od ponad 20 lat doradza firmom w budowaniu warto\u015bci wok\u00f3\u0142 nowych technologii. Konsultant, trener, praktyk i m\u00f3wca TEDx. Szkoli mened\u017cer\u00f3w i projektuje oraz wdra\u017ca procesy adopcji innowacji i narz\u0119dzi opartych o generatywn\u0105 AI."}],"_links":{"self":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/17708","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/users\/252"}],"replies":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/comments?post=17708"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/17708\/revisions"}],"predecessor-version":[{"id":17709,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/17708\/revisions\/17709"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/17666"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=17708"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=17708"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=17708"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=17708"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=17708"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=17708"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}