11 Months in 2025: AI Adoption — FOMO, Hype, or a New Paradigm?
By Ilya Ilienko, MBA, CMA, CPA, CFO
History and Parallels
Let's not ignore current realities and rush to buy in. Time constraint is one of the most well-recognized manipulation tools and sales tactics out there. Create urgency, scare people into making a decision and the herd mentality follows. Some examples of similar FOMO events have included the: 2000 Dot-Com bubble, 2008 real estate bubble, hyped IPOs, similar FOMO-driven crypto coin launches, and so on.
The time will come, but we must recognize what AI is today — and what it is not yet.
Two Sides of the Same AI Coin
Side 1: AI Bonanza!
Companies are plastering “AI-powered” or “intelligent” on every product and process. In the news you hear nothing but “It’s not hope. It’s now. It’s already happening!”. (Top Bank Says AI Adoption Already Paying-Off) “88% of organizations report using AI in at least one business function” (The State of AI)
In 2024, 95% of SMEs in Singapore adopted at least one digital solution, with 97% using at least one sector-specific digital solution. (Singapore Digital Economy Report 2025) HubSpot is even more rosy: 97% of businesses believe AI will transform their operations, and 95% of all communications are projected to utilize AI by end of 2025. (The State of AI In Business and Sales)
On the one hand, AI is an unstoppable force.
Side 2: AI Bust!
A quick “scratch and sniff” test — how true are these claims? Many of these companies are marketing and hyping up their own products — which makes sense, they have skin in the game. However, the story shifts when we look at actual results.
Let me counter the hype with a few data points. Several months ago, MIT published a report that states that 95% of AI companies are failing. (MIT report: 95% of Gen AI Pilots at Companies Fail). It's interesting to see that McKinsey — the same source cited earlier — reports that — 8 out of 10 companies see no revenue gains from AI adoption. (Over 80% of Companies Embracing AI See No Real Gains) The report illustrates that 80% of these projects do not produce any material result on investment.
If we scratch the surface we see not just somewhat opposing views, but mostly different angles: aspirational and promising VS. hard industry reality and ROI. So, if the statistics paint such an opposing picture, what does AI adoption really look like in practice? That brings us to how AI is deployed.
Intro to AI task automation, and . . .
How much of an edge will your company lose if your competitor automates non-core peripheral tasks, such as communication, design, and research? AI is a double-edged sword. A high-quality human approach takes time and may cost more money, but it can be relied on more than a black-box, algorithm-driven LLM. A trustworthy human doesn’t need the same level of oversight and doesn’t have to be reviewed and steered after every single action.
Even when AI tools work as intended, the way they’re implemented can create a new kind of complexity: fragmentation.
Fragmentation and an HR Example
A seasoned HR professional can handle end-to-end tasks quickly and can think outside of the box. AI automation has to be thought out, implemented, monitored, and tweaked or refined by a person — and now this person has to be fluent in both HR and AI implementation and oversight, whereas before you just needed a seasoned HR expert. Additionally, most companies are selling automations piecemeal. As in our HR example:
- 1st AI employee to research candidates which requires ongoing human feedback and throughput
- 2nd AI to reach out and carry out initial generic communication (pre-qualifiers and candidate information collection), until a human has to step in for oversight again, and OK the process to the next AI
- 3rd AI in line to handle calendar scheduling. If there is a quirk, or a non-standard request or need, a human has to step in
- 4th AI conducts the initial interview screening from the video recorded interview responses (sentiment analysis and flagging issues for human review)
- 5th AI to generate personalized offer letters based on role, compensation ranges, and candidate particulars (pending another final human approval)
- 6th AI to initiate background checks and verify results (escalating discrepancies to HR yet again)
Fragmented processes similar to these, don’t only require constant “pick-up and put-down” reviews and green-light decisions from human professionals, but they also invite opportunities for missteps along the joints. What is easier — have 1 HR professional handle A-Z of hiring at an SMB, or implement, tweak, and oversee 4 or more loosely connected processes? And this is just an HR example, how about marketing, legal, procurement, financial reporting . . . AIs can multiply exponentially. Think of your traditional accounting department — it has AP, AR, treasury, reconciliation, reporting, and period close. This is not a few super smart AI employees — this is dozens of specific AI task automations. Beyond process complexity, AI also changes how businesses represent themselves to the humans they interact with — customers, vendors, candidates, and employees. That’s where empathy and risk intersect.
Empathy and Risk
With these types of infrastructure architecture, upstream and downstream cascade effects increase exponentially. These systems often connect different platforms, like CRMs, ERPs, and cloud solutions, making it hard to identify the source of the issue until they’ve compounded (more on that later). Further, client, customer, vendor, and prospective employee facing processes — these folks are able to tell when they are dealing with AI. AI doesn’t think outside the box, it can’t reason as well through non-quantitative factors, or take soft skills into consideration, it also cracks when faced with non-routine requests. This doesn’t even touch on the human elements — understanding, compassion, EQ, patience, resilience, and other qualities AI can’t reproduce. It doesn’t care about your vendors, clients, candidates, or employees, it’s a transactional machine.
If a customer service representative asks, “How are you today?” and the customer replies, “Not great — I had to put my dog down this morning,” the human response will differ profoundly from AI’s. Brainstorm the progression of the conversation from there if AI VS. a human is speaking to the person who just lost their dog. Sympathy and empathy go a long way, and people want to be understood. The cracks will become more apparent as we advance further down to automation, less oversight, and more Agent-to-Agent communication. Automated AI agents don’t wait for human checkpoints, they run 24/7. Havoc can be wreaked overnight, and a small error will snowball fast. This isn’t the future, this is now. (10 famous AI disasters | CIO) From profane and racist chat-bots acting as customer reps, to demographic biases in HR, credit, and judicial systems, together with new and hoax propagating misinformation incidents.
In Conclusion For This Go Go Go World –
Friends — don’t let your guard down — let’s continue to think critically and not allow our emotions to sway us. Ask those important questions — how, why, where, and when? Let’s maintain our professional skepticism, keep investigating, and most importantly, remain patient. Thoughtful, steady, and observant.
“Slow and steady wins the race.”
— Aesop’s Fables — The Tortoise and the Hare.