New research coordinated by the European Broadcasting Union (EBU) and led by the BBC found that AI assistants misrepresented news content in nearly 45% of tested cases.
Similarly, a study by OpenAI study found that only one in four ChatGPT answers is correct.
So, does this mean we have to fact-check every single AI output?
Not necessarily.
This week’s newsletter explores how to get measurable ROI from AI without wasting hours on validation or risking your reputation in the process.
? Understanding AI ROI
AI delivers low ROI when the time spent verifying outputs, combined with legal and reputational risks, outweighs business benefits.
To boost AI ROI, focus on reducing validation time and minimizing exposure. The most effective steps are to:
- Invest in AI training for your team.
- Establish strong governance and usage policies.
- Test use cases before full rollout.
? High-ROI, Low-Oversight AI Use Cases
? How Top Tech CEOs Use AI
According to CNBC’s article CEO Sam Altman: How I use AI in my own everyday life—it’s great for ‘boring’ tasks:
- Sam Altman (OpenAI) uses AI for email summaries and simple tasks.
- Jensen Huang (Nvidia) relies on chatbots for drafting.
- Satya Nadella (Microsoft) uses AI to organize his inbox.
? How Google Thinks You Should Use AI
Google’s Unleash the Power of AI for Your Small Business highlights AI’s real value in freeing time and simplifying daily work.
It recommends applying AI in five key areas: sales and marketing, creative writing, customer service, operations, and bookkeeping.
? How AI Is Used at Perplexity
According to Perplexity at Work, the most impactful AI use cases focus on:
- Blocking distractions
- Scaling individual performance
- Delivering measurable results
Employees use AI to automate admin work, synthesize information, and drive results in areas like research, business development, and project delivery.
? How OpenAI Recommends Using AI
In Identifying and Scaling AI Use Cases, OpenAI notes that success comes from practical, everyday applications.
The most effective use cases fall into six “primitive” categories: content creation, research, coding, data analysis, ideation and strategy, and automation.
? Natalia’s “Low-Hanging Fruit” AI Use Cases
In our experience, AI shines in low-risk, low-oversight categories such as:
✅ Text & Speech Processing: summarizing, transcribing, and polishing content.
✅ Brainstorming: expanding ideas and creating outlines.
✅ Repetitive Tasks: automating reports, schedules, and tracking.
✅ Visuals: generating mockups, graphics, or short clips.
⚠️ Low-ROI, High-Oversight Use Cases
According to McKinsey’s Superagency in the Workplace, only 1% of enterprises have reached mature AI adoption.
From fake legal citations to data leaks, 2025 was full of AI missteps, reminding us that automation can cost more than it saves.
Avoid using AI for:
- High-risk or highly complex tasks.
- Sensitive customer interactions.
- Projects where accuracy is difficult to verify.
? Key Takeaways
1️⃣ Start simple. Automate low-stakes, low-risk tasks first.
2️⃣ Go deep, not wide. Focus on a few valuable use cases, not endless prompts.
3️⃣ Build strong foundations. Use small wins to prepare for larger-scale automation.
? Find more on this topic in the full article: A Practical Guide to AI ROI
? Your Turn
In what areas has AI worked well for you without much human oversight?
I’d love to hear from you — reach out at natalia@nataliabrattan.com.
Talk soon,
Natalia