Generative AI, or GenAI for short, is now on everyone’s minds. Whether you consider software a world-eating monster, as tech billionaire Mark Andreessen already proclaimed in 2011 ,or in association with a job apocalypse, as some of the most prominent AI thought leaders predict. The NVIDIA CEO Jen-Suan Huangsaid at a summit in Dubai in early 2024, that even at this early stage of the AI revolution, software development is no longer an important capability. Instead, one should focus on areas such as biology, education or agriculture.
But the more time passes, the clearer it becomes that a differentiated view of the possible bubble is necessary. Even Sam Altman, CEO of OpenAI, announced recently:"Someone will lose incredible amounts of money." Even if one ignores the fact that such a statement puts a considerable damper on the stock prices of OpenAI’s – mostly listed – competitors and one could assume a business interest here, there are further signs that the AI revolution may not be as world-devouring as one would suspect at first glance into the hype bubble.
Investments without outcomes, data breaches and lack of experience
According to the August 2025 Report published According to the MIT Media Labs NANDA report, 95 percent of investment in generative AI would bring absolutely no measurable benefit – silch, zero, nothing. Instead, it seems that only the sellers of the shovels of this AI gold rush are earning a decent amount of money.
Many users also seem to trust the AI models too much and – IMHO – build unhealthy relationships with the ChatBots, one example of that being the emotional outcry during OpenAI’s model change from the 4 series to GTP-5. The new model was simply too emotionless for many users. This is quite worrying considering the recent privacy breaches at OpenAI and xAI. If a user shared a conversation via link, they were directly accesible through search engines.In addition, ChatGPT advises its usersthat their supposed private conversations with the chat bot could be used as evidence in a New York Times lawsuit. After the tragic events of recent months, such as the suicide of a teenager in the US, OpenAI is now even starting to scan conversations with the chat bot for potential – let’s say – problems.. Of course, this applies not only to personal conversations, but also company secrets, or simply API keys in the software code.
Vibe coding: the death of the junior developer?
Anyone who deals with the topic of vibe coding could get the impression that all this is in the end is not that important. The new technology brings too many advantages. Similar to NoCode/LowCode, this approach allows everyone to be a software developer. Of course, it is welcome if the entry into software development is accessible to everyone who is interested in it and hurdles are gradually reduced. The fact that there will soon be no need for software developers is still difficult to imagine despite all the hype around vibe coding. It is much more of a long-term danger. When CEOs like Satya Nadella announcethat around 20 to 30 percent of the software in the companie's own repositories was written by AI: You cannot compare the resources of an industry giant with a vibe-coded project.
The software developer David DeGraw, for example, has published a blog post stating that it should be noted that companies rely on their senior developers to verify the accuracy of vibe-coded commits of inexperienced colleagues. I too have made similar expiriences in my career.Like the team of Answer.AI. AI-generated answers often lead developers into a "rabbit hole" consisting of devised functions and complex approaches. This leads to a lot of wasted time in the development – if only I had read the documentation once.
All that glisters is not gold
Should you as a prospective freelance software developer completely ignore the trend and only refer to books? Of course not, but the question arises as to how – especially in the context of customer projects – one can possibly deal with the use and the associated risks of AI in software development. So how can data protection risks or cost explosions be avoided and how do you deal respectfully with the time of your colleagues? I will explore this in the next parts of this series.
