Ai Project Failures - Why So Many?

When you think how much time, effort and money has gone into introducing artificial intelligence, how many of these projects have failed? [Sources: 4]
    
Research has shown that there are so many ways to fail, and there seem to be several reasons why AI projects have failed, including lack of funding, poor management, poor design, or just bad luck. But before we go into the reasons why some of these AI projects have failed, let us look at some examples of how some might fail. There are a number of factors in the process of actually starting a project, such as the quality of the project itself, the amount of resources available to the team and so on, and other factors. [Sources: 4]
    
Note: This opinion does not reflect the current or former employer, but is the opinion of the author and his personal experience. Below is a list of why AI projects have failed and why they are of no use. This project was started after the company hired a few data scientists to build their model in Python (r), only to quickly discover that there was a difference in thinking between data scientist and engineer. [Sources: 5]
    
In the short term, this problem is often solved by the machine learning team taking over the code, rewriting it, and following the standard dev / ops process to scale and deploy the application. In analytics and AI, the process of continuous improvement would apply technical features such as modeling estimates, refining the measures associated with model output, quantifying the goodwill generated by the models, etc. [Sources: 5, 9]
    
But often these projects look good on paper, but do not deliver meaningful results when it comes to actually implementing these strategies and delivering results such as a return on investment. This is why so many analytical and AI projects fail, what we call the staircase of failure. For machine learning or AI to succeed, it must overcome many of the same problems that many software projects have failed, many data warehousing projects that have failed, and many complex projects that have failed. [Sources: 2, 9]
    
Seventy percent of companies report minimal impact from artificial intelligence projects, but that's not all. IBM's Watson AI, which is used in oncology, was cancelled after the company was due to spend $62 million. She is said to have made false recommendations for cancer treatments, according to a report in The New York Times and The Washington Post. [Sources: 2, 6]
    
There are a number of reasons for this, including the need for organisations to adapt to new working practices, but the most important factor is poor data. This includes everything from the lack of data to the poor quality of the data itself, to poor communication between employees. [Sources: 6]
    
To use the comparison that AI is like icing on the cake, data cakes are without data, and AI would be the icing on the cake. [Sources: 6]
    
As a data scientist, not everyone is familiar with artificial intelligence and machine learning, but data scientists say there is a huge gap between public perception of AI projects and reality. Data scientists also have a lot of data on how 85 percent of the AI project is not delivering on its intended promises. [Sources: 0]
    
One of the challenges of adopting artificial intelligence is that management may not recognize the value of this emerging technology and be willing to invest as such. AI is still seen as an expensive tool for measuring and maintaining, and there are no resources to educate users about how the two technologies work. [Sources: 0, 3, 4]
    
It is not uncommon for companies to point the finger at the failure of AI projects as the cause of their failure, and the blame is backed up by a number of examples of global organizations using AI solutions. Create success and failure criteria, track appropriate metrics and ROI, prepare teams to work with the system and prepare them for the right approach, which includes understanding the business problems that artificial intelligence can solve, and develop a data strategy. [Sources: 1, 3]
    
Most organizations report the failure of their AI projects, with a quarter of them reporting a failure rate of 50%. A recent survey of global organizations that already deploy AI solutions found that only 25% have developed an enterprise-wide AI strategy, despite 2 / 3 emphasizing an AI First culture and 3 / 1 / 2 prioritizing AI. [Sources: 1, 8]
    
Too often, AI has not delivered the positive impact that businesses really want from technology, such as fewer customers lost to brain drain. Some firms blame the main blockers that are internally hampering the adoption of artificial intelligence: lack of funding, technical challenges, and a lack of awareness of the benefits. [Sources: 7, 8]
    
More than 60% of companies reported at least one failure of AI projects last year, and a full quarter reported a failure rate of 50%. The two main reasons for the failure were the lack of skilled workers and unrealistic expectations. Respondents reported an average of 1.5 failures per 100,000 AI projects. Despite some bumps on the road, there is little doubt that AI is an unstoppable force, especially in business. Nearly 50 percent said they had created a formal framework to encourage the use of AI in their business, up from 30 percent in 2015. [Sources: 7]
    

Sources:
    
[0]: https://www.gizbot.com/news/data-scientists-claim-85-ai-projects-fails-what-they-promise-063672.html
    
[1]: https://www.iunera.com/kraken/big-data-science-strategy/3-unignorable-reasons-big-data-and-ai-projects-fail/
    
[2]: https://www.mediapost.com/publications/article/337368/why-85-of-artificial-intelligence-projects-fail.html
    
[3]: https://neoteric.eu/blog/ai-projects-fail-heres-why/
    
[4]: https://medium.com/@NeotericEU/the-single-most-important-reason-why-ai-projects-fail-68c8c1f72dee
    
[5]: https://towardsdatascience.com/five-hypotheses-as-to-why-artificial-intelligence-and-machine-learning-projects-fail-7c6b2c456d41
    
[6]: https://www.itproportal.com/features/why-ai-projects-fail/
    
[7]: https://venturebeat.com/2019/07/08/idc-for-1-in-4-companies-half-of-all-ai-projects-fail/
    
[8]: https://www.fastcompany.com/90449015/this-much-hyped-technology-is-failing-businesses-heres-why
    
[9]: https://analyticstrategy.com/tag/ai-project-failures/
    

All Posts
×

Almost done…

We just sent you an email. Please click the link in the email to confirm your subscription!

OK