The contribution of millions: Why many companies on AI failed first appeared at the online magazine Basic Thinking. You can start the day well every morning via our newsletter update.
Artificial intelligence has found its way into numerous companies. But a new co-study shows that almost all AI pilot projects fail in companies-and that millions are burned.
AI tools represent strong development fields for numerous companies. In 2024 alone 19.75 percent of companies in Germany artificial intelligence. In the meantime, the number is likely to have increased, as it was only 11.6 percent a year earlier.
In companies, AI is primarily used in customer contact, like 86 percent of the respondents of one Bitkom survey from 2025 information. In 47 percent of companies, generative AI is used in marketing and communication.
But although generative artificial intelligence is promising for companies, a large part of the AI pilot projects fails. That shows one New investigation of the Nanda initiative of MITfor which 150 managers were interviewed, among other things.
AI projects do not prevail in companies
For their report, the co-researchers examined the status of the AI in the economy. The investigation shows that only five percent of AI pilot projects in companies can achieve a quick increase in sales.
The majority, on the other hand,, i.e. the remaining 95 percent, has hardly or no measurable influence on the profit and loss account of the respective company. This is mainly due to the fact that numerous projects come to a standstill relatively quickly.
The researchers have conducted 150 interviews with managers for their investigation and carried out a survey of 350 employees. In addition, 300 public AI implementations were analyzed.
However, the problem with unsuccessful AI projects is not the quality of the models. Rather, there is a “learning gap” that is mainly created by poor integration in companies.
In addition, the resources in the companies are wrongly aligned. Because, according to the results, these spend more than half of their budget for generative AI for sales and marketing tools. The biggest return, on the other hand, can be achieved in the back office automation.
So the implementation can succeed
But while numerous AI projects fail in companies, the scientists were also able to identify successful models. “Some pilot projects of large companies and younger start-ups are really successful with generative AI,” explains Aditya Challapally, main author of the study, Compared to the business magazine Fortune.
Start-ups, whose bosses are between 19 and 20 years old, have “achieved a sales jump from zero to $ 20 million within one year,” continued Challapally. “This is because they concentrate on a weak point, implement it well and work cleverly with companies that use their tools.”
Among other things, the way of how companies use AI is crucial. Accordingly, the success rate increases when AI tools are bought from specialized providers to 67 percent. In contrast, only a third is successful in internal developments.
Also interesting:
- Smartphone-KI: Apple, Google and Samsung in comparison
- Electricity consumption of GPT-5: As high as from 1.5 million households-per day
- Chatbots: Why you can’t trust AI
- Artificial intelligence: Why AI scares many people
The contribution of millions: Why many companies on AI failed first appeared on Basic Thinking. Follow us too Google News and Flipboard Or subscribe to our update newsletter.
As a Tech Industry expert, I believe that there are several reasons why many companies fail on AI implementation. One of the key reasons is the lack of understanding and expertise in AI technology among decision-makers and leaders within the organization. Without a deep understanding of how AI works and its potential applications, companies may struggle to effectively integrate AI into their operations.
Additionally, companies often fail to properly define their AI goals and objectives, leading to unclear expectations and outcomes. Without a clear strategy and roadmap for AI implementation, companies may struggle to measure the success of their AI initiatives and make informed decisions on how to improve and iterate on their AI solutions.
Another common pitfall is the lack of quality data and infrastructure needed to support AI projects. AI algorithms rely heavily on high-quality, relevant data to produce accurate and meaningful insights. Companies that do not have access to clean, comprehensive data sets may struggle to effectively train and deploy AI models.
Moreover, companies may also face challenges in attracting and retaining top AI talent. The demand for AI professionals is high, and companies that do not have a strong team of AI experts may struggle to effectively implement and manage AI projects.
Overall, successful AI implementation requires a combination of technical expertise, strategic planning, data quality, and talent management. Companies that fail to address these key factors are at risk of falling behind in the rapidly evolving AI landscape.
Credits