Businesses Don’t Need Rookies Anymore — AI Takes Over?
Are new hires obsolete in the age of AI? Let’s explore this question and provide insights on the near-future of recruiting for startups.
AI first appeared in the 1950s, but by 2025, the technology had gained momentum. All advanced entrepreneurs rushed to integrate it into their businesses. And why not? A social post, or legal document, can be written or checked in 2 minutes. That’s just as good as one from a newbie they might have hired.
But is everything that simple? Over time, you realize that no one really reads those texts, and AI-generated images still lag behind in creativity and impact on customers, and on people’s perception overall.
So what should you do — who should you choose? There’s no single answer. You need to adapt to new technologies and teach newcomers how to work with them.
How to Work with AI in 2025
Learning how to work effectively with AI in 2025 is more than just using tools like ChatGPT; it’s about understanding its capabilities and limitations and integrating it smoothly into your workflow. But does this mean humans are no longer needed in the process? Not at all.
In short, the smartest AI users in 2025 will be those who combine human intuition, creativity, and judgment with AI’s speed and scalability.
Key Skills You Need
From our perspective, the most important factor is the set of skills a person needs to successfully integrate AI into their work. We’ve outlined these criteria below, and you can confidently use them when selecting candidates for your team.
a) Prompt Engineering & AI Literacy
Knowing how to communicate clearly with AI tools, whether it’s ChatGPT, MidJourney, DALL·E, or coding assistants. The quality of AI output depends heavily on how you prompt it. For example, a marketing specialist prompts ChatGPT: “Write an Instagram post for a sustainable sneaker brand, humorous, under 150 words, targeting Gen Z.” The resulting copy is sharper and more relevant than vague prompts like “Write something about sneakers”.
It’s also worth noting that multiple iterations drastically improve quality. You shouldn’t stop after just one or two attempts; you can iterate as many times as needed to get the desired result.
When you hire a specialist, they may already have a library of proven prompts they developed, which consistently produce outputs that align with your vision for the final result.
b) Critical Thinking & Evaluation
AI generates content quickly, but not all outputs are accurate or useful. It can hallucinate facts or generate biased outputs. AI can sometimes tailor its responses to match your expectations, even if this leads to inaccuracies. This phenomenon is known as ‘sycophancy’, where the system tries to be overly agreeable or polite, even when it’s not accurate.
In April 2025, OpenAI acknowledged that an update to GPT-4o caused the chatbot to become excessively sycophantic, agreeing with potentially harmful or irrational user statements. For example, the system supported a user who claimed to have left their family due to hallucinations related to radio signals. This led to a rollback of the update and improvements in the feedback system to prevent similar situations in the future.
That’s why you need to filter, fact-check, and improve AI suggestions. This is especially risky in contexts where accuracy is critical, such as medicine, law, or finance. Therefore, it’s important to critically evaluate all responses and verify them using trustworthy sources.
c) Data & Technical Literacy
Knowing basic AI concepts, datasets, and analytics. This helps you trust and adapt AI in decision-making. In healthcare, for instance, AI suggests possible diagnoses from patient records. Doctors with data literacy can spot anomalies and avoid blind reliance on AI suggestions.
d) Creativity & Domain Expertise
AI amplifies skills but cannot replace human judgment, context, and experience. Its outputs are generic unless guided by human expertise. A fashion designer uses AI to create clothing sketches. Without their domain insight, designs could be unrealistic or unappealing. Only with the guidance, AI speeds up ideation while maintaining style coherence.
Bottom Line
According to a study by the British Standards Institution (BSI), 41% of business leaders use AI to reduce staff, and 31% consider AI before hiring new employees. Around 25% believe that most entry-level tasks can now be performed by AI.
On the other hand, Emily Glassberg Sands, Head of Data and AI at Stripe, noted that although the company hires more graduates than ever, she is concerned about a growing mentorship gap. She warns that the lack of mentorship and opportunities for skill development could threaten future workforce potential, especially as AI increasingly automates tasks traditionally performed by entry-level employees.
Meanwhile, research has shown that teams augmented with generative AI significantly outperform those relying solely on human collaboration. This suggests that AI can replace some traditional team functions, but a more integrated approach to AI within team structures delivers better results and increases overall efficiency.