In the recent Harvard Business Review article “AI-Generated ‘Workslop’ Is Destroying Productivity,” the authors define AI-slop as the growing volume of low-value output and operational overhead created by AI tools in the workplace.
The statistics are staggering: 95% of organizations report no measurable ROI from their AI investments.
From our work with more than 100 SMEs, we’ve identified three major mistakes that small businesses often make when adopting AI.
These missteps frequently lead to the problem of AI-slop.
Mistake #1: Giving AI More Responsibility Than It Can Handle
One of the most common mistakes small businesses make with AI is assigning it to critical, high-stakes, or highly specialized tasks. It may sound logical. After all, AI adoption advice often starts with “find a strong business case.” But when the stakes are high, the risks multiply.
AI still makes factual errors, or what experts call hallucinations. Research shows that one in four ChatGPT responses is inaccurate. It means you could spend more time double-checking AI’s work than if you’d completed the task manually. That’s why the smartest AI strategies begin with low-risk, low-complexity tasks where human oversight can remain minimal.
OpenAI CEO Sam Altman highlighted this at a Harvard University Fireside Chat. He described a tutoring company that could either spend a lot of effort refining AI to just barely support 8th-grade students today. Instead, the company can deploy AI effectively for 6th-grade students with minimal supervision, and then gradually scale as the technology improves.
Altman pointed out that many entrepreneurs overestimate AI’s current limitations but underestimate its rapid improvement.
It’s telling that Altman himself primarily uses AI mostly for low-stakes tasks such as managing email, summarizing documents, and automating routine tasks. Similarly, Jensen Huang, CEO of Nvidia, uses AI to generate first drafts of documents.
Mistake #2: Chasing Prompts Instead of Building Processes
The internet is overflowing with endless lists of “100 AI prompts for every situation.” It’s tempting to spend hours experimenting with them, but that’s the second biggest mistake SMEs make when adopting AI: focusing on prompts rather than purposeful business outcomes.
As Alexandr Wang, founder of Scale AI and one of the youngest self-made billionaires in tech, explains:
“It’s not what humans perceive to be the simplest task that AI will automate first, but rather where we have the most data.”
AI isn’t powerful because of the prompt. It is powerful because of the data it can learn from and act upon. In fact, studies show that 95% of users still consider traditional search engines more trustworthy than AI tools. The main competitive advantage of AI is in using your data to do execute business tasks for you.
To unlock real AI value, shift your focus from experimentation to execution. Identify business areas that are:
- Data-rich, but non-sensitive and non-confidential
- Repetitive rather than one-off tasks
- Predictable in structure and outcome
Great examples include invoicing, customer feedback analysis, or historical sales forecasting. These processes provide the consistent data patterns AI thrives on.
Once you’ve chosen a focus area, customize your AI tool and upload your standard operating procedures, set clear instructions, and define the quality standards you expect. Then, test, refine, and iterate.
Mistake #3: Failing to Provide AI Governance
Many companies, including small businesses, are rolling out AI tools like Microsoft Copilot or ChatGPT to their teams with the expectation that employees will “figure it out.” The hard truth? They often don’t.
Without proper governance, AI adoption quickly unravels. Employees either avoid using the technology altogether, fearing it could replace them, or they use it ineffectively and unsafely unintentionally exposing sensitive information or generating unreliable results.
According to the Enterprise AI and SaaS Data Security Report by AI and browser security company LayerX, between 62% and 77% of employees leak sensitive business data to AI tools, often without realizing it. For instance, if your team uses ChatGPT to draft emails or summarize reports, they could be sharing customer or company information and putting your business at serious risk.
To prevent this, business owners must establish strong AI governance from the start. At a minimum, this means:
- Defining clear purposes and business cases for AI use.
- Setting strict rules about what data can and cannot be shared.
- Creating role-based access to control who uses AI tools and how.
- Enforcing multi-factor authentication wherever possible.
- Requiring work emails for all AI-related accounts.
While AI regulations in Canada are still evolving, SMEs must comply with existing laws governing data protection and privacy, including PIPEDA and relevant provincial acts. Business owners should also review AI vendors’ privacy policies regularly to ensure they align with acceptable security standards and ethical practices.
Key takeaways:
- Start with what AI can handle, not what you want to automate.
Focus on tasks that AI performs reliably with minimal human oversight, and avoid assigning it responsibilities beyond its capabilities. - Go beyond prompts and build AI processes and workflows.
Identify business areas rich in non-sensitive, structured data, and develop tailored AI solutions supported by clear instructions and repeatable processes. - Establish strong AI governance from day one.
Create rules and safeguards that ensure safe, ethical, and effective use of AI within your organization. Regularly review the privacy and security policies of AI vendors to stay compliant and protect your business.
Final Thought
For Canadian SMEs, the real opportunity isn’t in adopting AI for the sake of innovation, it’s in adopting it strategically. Start small, stay safe, and scale thoughtfully. With the right balance of experimentation, structure, and oversight, AI can bring measurable and sustainable business results.
About the Author
Natalia Brattan is a Harvard-trained AI expert, consultant, and published author. She led technology audits for Cisco, advising hundreds of SMEs, and worked at BlackBerry specializing in audit and risk management.
