Appen, an AI data provider, released its 2024 state of AI report which reveals that many companies are struggling to power their AI systems because of countless high quality data. The report conducted a survey of 500 US IT experts and found out that there was a 17% increase in AI adoption in different companies this year. The report talked about several areas where the companies are experiencing problems with the adoption of AI.
The report states that even though there is a 17% increase in AI adoption in 2024, there have been some hurdles seen in data management. AI models need to be customized to work for different purposes in different enterprises but companies are somewhat struggling with using specific data for specific cases. Custom data collection is being used to train AI models by sourcing training data now.
The report also found out that a lot of companies are unable to reach AI deployment and the ones which become successful in doing so are seeing decline in ROI. There was a drop of 8.1% in completing AI deployment since 2021 while the ROI of companies with deployed AIs have declined by 9.4%. This drop is because of increasing complexity of AI models which is making it harder for companies to implement them.
Data accuracy and data quality has also declined by 9% since 2021. As AI models are getting more complex, their data is also getting complex. 86% of the companies try to update their AI models every quarter, but the data accuracy and quality keeps on declining with every update. 90% of the businesses are now taking help from third party sources to train their AI models. Data also keeps on getting bottlenecked, with 10% YoY increase. This is directly impacting the companies and their AI projects. Companies are planning long term strategies to tackle this issue like seeking strategic partnerships to understand and implement AI data.
Appen’s report also stated that even though many companies are shifting towards AI, the need for humans is still there. 80% said that human involvement is very important in AI machine learning as it can help develop AI systems that are high performing and relevant.
Read next: AI’s Limitations in Storytelling: What This Study Reveals About Reader Preferences
The report states that even though there is a 17% increase in AI adoption in 2024, there have been some hurdles seen in data management. AI models need to be customized to work for different purposes in different enterprises but companies are somewhat struggling with using specific data for specific cases. Custom data collection is being used to train AI models by sourcing training data now.
The report also found out that a lot of companies are unable to reach AI deployment and the ones which become successful in doing so are seeing decline in ROI. There was a drop of 8.1% in completing AI deployment since 2021 while the ROI of companies with deployed AIs have declined by 9.4%. This drop is because of increasing complexity of AI models which is making it harder for companies to implement them.
Data accuracy and data quality has also declined by 9% since 2021. As AI models are getting more complex, their data is also getting complex. 86% of the companies try to update their AI models every quarter, but the data accuracy and quality keeps on declining with every update. 90% of the businesses are now taking help from third party sources to train their AI models. Data also keeps on getting bottlenecked, with 10% YoY increase. This is directly impacting the companies and their AI projects. Companies are planning long term strategies to tackle this issue like seeking strategic partnerships to understand and implement AI data.
Appen’s report also stated that even though many companies are shifting towards AI, the need for humans is still there. 80% said that human involvement is very important in AI machine learning as it can help develop AI systems that are high performing and relevant.
Read next: AI’s Limitations in Storytelling: What This Study Reveals About Reader Preferences
