What are Human Skills?
In 2014, Amazon's machine learning team had a vision to build an AI that could screen thousands of resumes and identify top talent automatically. By 2015, they realised their algorithm had learned something they never intended [1].
It taught itself that men were better.
The system penalised resumes with "women" in them. Amazon's engineers spotted the bias and tried to fix it. They edited the code. They neutralised the problematic terms.
But here's the terrifying part: they couldn't guarantee the algorithm wouldn't find new ways to discriminate.
AI saw patterns in data. But it couldn't ask whether it should be learning from a biased past. AI can't question its own assumptions. Humans had to step in to prevent the damage.
Humans WITH AI
This isn't Amazon's failure, but a window into AI's fundamental limitation. And it reveals an opportunity: AI-Human collaboration doesn't diminish human value. It amplifies what makes us most human.
Amazon invested millions in their AI. But humans caught the bias. Humans tried to fix it. Yet they couldn't trust AI to stop discriminating because machines can't question whether their patterns are right.
We shouldn't be asking whether AI will replace humans. Instead, we should ask: How do we humanise the way we work with AI to create sustainable, human-centred growth?
What Makes Human-AI Collaboration Work
There are three intelligences that AI will always struggle with. These aren't soft skills or nice-to-haves; they are strategic capabilities that separate leaders from followers. These are our human skills.
Pillar 1: Judgmental Intelligence
The power of human judgment
Toyota popularised Kaizen (continuous improvement). Now they're championing Jidoka (automation with a human touch), which gives machines the ability to detect problems while keeping humans in charge of judgment about what to do next [2].
With Toyota's AI-Human collaboration platform, factory workers with zero coding experience build their own machine learning models through simple interactions [3]. The software monitors for defects, predicts equipment failures, and flags abnormalities. When parameters trigger a stop, a worker must decide if there's a defect or if the machine is adapting appropriately.
At Toyota's Motomachi Plant, the software flagged an irregular adhesive application and stopped the line [3].
Kenji, a veteran worker, asked questions AI couldn't:
- Is this irregular, or optimal for this custom model?
- Did humidity affect the adhesive?
- Should we adjust the AI or the process?
After analysing the conditions, Kenji determined that the application was better adapted to the environment than the AI's standard. He updated the work instructions and trained the AI to recognise this as acceptable [3].
Why does this matter: While AI sees patterns and handles more routine analysis, the ability to question AI’s recommendations becomes a leadership superpower. In a world where every competitor has access to the same AI tools, your judgment about when and how to use them becomes your competitive advantage.
Pillar 2: Relational Intelligence
The art of authentic connection
AI chatbots are everywhere now, even in mental health. Modern chatbots are available 24/7 and can analyse text for distress signals and suggest coping strategies based on cognitive behavioural therapy. These systems log symptoms and flag risks, but when emotional nuance matters, a human therapist must decide what a patient truly needs.
A 2025 Stanford University study revealed that while AI therapy chatbots could recognise keywords associated with depression and provide evidence-based responses, patients consistently reported feeling heard but not understood [4].
Here's a scenario:
Patient says: "I'm fine."
AI chatbot: Logged "positive affect," moved to the next assessment question, and provided general coping advice.
A human therapist noticed what AI couldn't:
- Is the flat tone signalling resignation rather than wellness?
- Did the pause before "fine" indicate something hidden?
- Should I push deeper or give them space right now?
The therapist responded: "I don't think you're fine. And I think you're tired of people not seeing that."
The patient broke down, not because of advice, but because someone finally saw through the mask.
Why does this matter: Trust is the foundation of every high-value relationship, be it with customers, patients, employees or your partner. This is especially true in an industry that requires the human touch. For instance, in healthcare, when empathy is missing, patients withhold information and do not adhere to treatments. As AI commoditises information and efficiency, your ability to build genuine human connections becomes the ultimate differentiator that competitors can't replicate.
Pillar 3: Adaptation Intelligence
The wisdom of when and how
In 2017, a Harvard Business School study by Edelman, Luca, and Svirsky revealed that Airbnb users with African-American-sounding names were 16% less likely to receive booking approvals than users with white-sounding names [5]. This discrimination didn't stem from the algorithm, but it emerged from human hosts making biased decisions that the algorithm amplified by learning from historical patterns.
In one case, the AI flagged a host who rejected three consecutive bookings from guests with Black-sounding names. A human specialist investigated beyond the data and discovered the host had legitimate concerns about group size violations at their small apartment. The AI only analysed the demographic patterns, not the contextual reasoning behind the decisions. The AI couldn't distinguish the context from racial bias
Human specialists asked questions AI couldn't:
- Is this racial discrimination, or do these bookings share other characteristics (group size, last-minute requests, incomplete profiles)?
- How does cultural context affect what constitutes suspicious behaviour versus legitimate caution?
- Should we remove the host, provide education, or is there missing context we need to understand?
Why does this matter: context is invisible to algorithms but obvious to humans. Airbnb's discrimination challenges show that statistical patterns without cultural understanding create legal, ethical, and brand risks. As businesses operate across diverse markets and contexts, the ability to read between the lines, understand cultural nuances, and make judgment calls about what's appropriate separates companies that scale globally from those that fail spectacularly. AI provides the pattern. Humans provide the wisdom about whether that pattern should guide action.
Human Skills Make AI Work
Notice what all three examples demonstrate:
These three human intelligences are proven capabilities, not theoretical concepts. Organizations like Toyota and Airbnb rely on AI-Human collaboration daily. However, most professionals haven't been trained to develop these intelligences systematically in their specific roles.
What You'll Receive Every Wednesday
That's where the Edge Series comes in: three focused newsletters each month to build this intelligence for your work:
The Selling Edge
How revenue teams use AI for research while building the authentic trust that actually closes deals.
The Voice Edge
How to partner with AI content tools without losing the authentic credibility that creates your human differentiator.
The Impact Edge
How customer success teams use AI to spot problems early, then apply human empathy and judgment to solve them
Subscribe now. While others fear AI, we're building the human skills that make AI collaboration work. Every week you wait, they gain ground.
You're not waiting. You're here.
Join our growing community.
Sources
[1] Reuters (2018). "Amazon scraps secret AI recruiting tool that showed bias against women." Jeffrey Dastin. Available at: https://www.reuters.com/article/us-amazon-com-jobs-automation-insight-idUSKCN1MK08G
[2] Toyota Production System. "Kaizen and Jidoka principles." Toyota Motor Corporation official documentation on lean manufacturing and automation with human touch. Available at: https://global.toyota/en/company/vision-and-philosophy/production-system/
[3] Toyota Case Study. "AI-Human Collaboration Platform at Motomachi Plant." Details on factory workers building machine learning models and quality control processes. Available at: https://news.microsoft.com/source/asia/features/toyota-is-deploying-ai-agents-to-harness-the-collective-wisdom-of-engineers-and-innovate-faster/
[4] Stanford University (2025). "Study on AI therapy chatbots and patient satisfaction: Recognising keywords versus understanding emotional nuance." Available at: https://news.stanford.edu/stories/2025/06/ai-mental-health-care-tools-dangers-risks
[5] Edelman, B. G., Luca, M., & Svirsky, D. (2017). "Racial Discrimination in the Sharing Economy: Evidence from a Field Experiment." American Economic Journal: Applied Economics, 9(2), 1-22. Harvard Business School Working Paper 16-069. Available at: https://www.library.hbs.edu/working-knowledge/airbnb-hosts-discriminate-against-african-american-guests
[6] Airbnb (2020). "Project Lighthouse: Fighting discrimination through AI-powered monitoring." Airbnb Newsroom and Trust & Safety reports. Available at: https://news.airbnb.com/2024-project-lighthouse-update/