Enterprise AI ambitions are at odds with distrust towards AI

Neil Raden Profile picture for user Neil Raden May 23, 2023
All we hear about is the massive transformative potential of AI. But on the ground, the reality of AI perceptions is different. How can organizations solve the data issues and other blockers that hold back innovate AI pursuits?

Robot model standing on dollar bills holding word ready? © yadali - Fotolia.com

Despite widespread recognition of the potential of artificial intelligence (AI), many organizations still need to be more open to fully embracing the technology. Senior IT and data science professionals acknowledge AI's significance, but most agree they must trust AI more before foregoing human-driven decision-making in many areas. This disparity indicates that while organizations understand the ambitious potential of AI, they are still hesitant to embrace it fully.

The enduring human element in decision-making within organizations may stem from underperforming AI models, reducing efficiency and certainty. Though people heavily involved in the decision-making process indicate that technology needs to be utilized to its full extent, they emphasize the need for organizations to ensure their data quality is up to scratch to leverage AI's benefits fully.

With organizations losing a proportion of global annual revenue because of underperforming AI programs that use low-quality data, it is crucial to ensure data quality. To fully embrace the power of AI, organizations must overcome their distrust and invest in the necessary infrastructure and talent to make it work for them.

The promise of artificial intelligence (AI) and machine learning (ML) to transform businesses and industries is well-documented. However, many organizations are still in the early stages of their AI journey, with much room for improvement.

The adoption of artificial intelligence (AI) has been on the rise across various industries in recent years. Organizations are investing more in AI-driven technologies, and the benefits are clear. From increasing efficiency to improving customer experience, AI has the potential to revolutionize the way businesses operate. However, despite the significant potential, organizations still need to be a considerable gap between organizations AI ambitions and the reality of their distrust towards AI.

There are multiple and multifaceted challenges to AI adoption, including:

  • The manual nature of data processes
  • Issues with data access and insight
  • Lack of time, skills, and budget, and ultimately:
  • A fundamental lack of priority given to AI.

Many organizations recognize that AI is the future and plan to increase investment accordingly. However, more trust in machine-led decision-making may be needed to overcome this.

Poor data management is the main obstacle to organizations' ability to leverage AI fully. This hinders even basic data analysis areas, and breeds technology distrust. To overcome this, organizations are beginning to realize that dismantling technical barriers to AI adoption will empower existing talent and make data AI-ready.

However, more than simply collecting vast amounts of data is required. The data must be usable and accessible to realize AI's benefits fully. Data scientists are more likely than senior IT decision-makers to recognize this, as they better understand what AI use looks like and how to achieve it.

One solution uses the modern data stack, which combines three core technologies - automated data integration, a cloud data warehouse or data lake, and a business intelligence or data visualization platform. By boosting their data's visibility, accuracy, and reliability, organizations are laying the foundation for a successful AI program and paving the way for more robust analytics capabilities altogether. However, the modern data stack is not without controversy, including the viability of federated data. See my recent piece, The modern stack has a messy data problem.

Another hurdle organizations face is using operational systems in Machine Learning (ML) projects, which varies across sectors. Organizations must ensure that their data is accurate, accessible, and formatted in a way that is easily used in ML projects for these projects to be successful. Media, retail and manufacturing companies are far ahead of, for example, the public sector, which is the least likely to use data from operational systems in their ML projects because they are the least likely to collect such data.

My take

A noticeable trend in the AI industry is the movement from generic applications of AI tooling to more targeted specific applications of the technology. For example, instead of building models from scratch to discover data in an organization's collections for duplicates, relationships, useful metadata, and automated replication and transformation, vendors offer all of these capabilities as applications with Ai behind the scenes.

Formerly, supply chain management software required organizations to build their models in rigid rules engines, which required constant tweaks and testing. New Ai-driven tools are available to sense and adjust to events in real time.

A grey colored placeholder image