Being able to implement artificial intelligence in an organization involves more than just having a basic understanding of the technology. According to a recent survey, most businesses and their IT departments are just not prepared to handle AI at this time. This is especially true for managers and executives who oversee the resources needed to make progress. In addition, the necessary solutions, skills, and tools are still lacking.
According to a poll conducted by SAS among 1,600 IT decision-makers, even heads of IT departments are still unaware of the potential effects of artificial intelligence. The majority of senior IT decision makers, or 9 out of 10, (93%) acknowledge that they are not entirely knowledgeable about generative AI (GenAI) or how it might affect business procedures.
It is imperative that executives are brought up to date. at the study, less than half (45%) of CIOs and slightly more than a third (36%) of CTOs said they were “extremely familiar” with the use of GenAI at their companies. Even worse, only 13% of chief digital officers acknowledge having a deep understanding of AI.
Things worsen: Just 4% of information systems or IT heads and 2% of IT managers or directors say they are quite familiar with AI.
Just 7% of organizations offer high-level training on AI governance and monitoring overall, while 15% more offer this kind of support for generative AI. This is crucial since 75% of respondents expressed worries about the security and privacy of their data when using GenAI in their businesses.
This implies that resolving the problems that could prevent the use of AI may require some time, in addition to extensive study and research. For instance, just 5% of them have a robust framework in place to assess the risk of bias and privacy in large language models. An further 42% of respondents are thinking about creating internal tools for detecting privacy risks, and 32% are thinking about creating internal tools for detecting bias.
Just 29% of generative AI solutions are continuously monitored by automation. Just 25% of companies routinely verify their AI output by manually.
According to the research’s co-authors, while the ideal GenAI investment presents obvious opportunities for efficiency and an improved customer experience, many firms identify strategic thinking gaps that are impeding successful adoption. According to our research, companies are using GenAI quickly without first putting in place sufficient governance frameworks. This could lead to major problems with quality and compliance down the road.
Another source of issues is the incorporation of AI into current systems and processes. According to the survey’s authors, “many companies struggle to integrate the technology with their existing tasks and tools.” Furthermore, nearly half (47%) of decision-makers state that they lack the necessary tools to apply GenAI.
The following are the main problems that businesses utilizing AI are encountering:
- Effective use of both public and proprietary datasets is a problem for 48% of respondents.
- 45 percent report a lack of appropriate tools.
- Of those who have used generative AI in practice, 42% say they are having trouble moving it from a conceptual stage.
- 39% report that their existing systems are not compatible with them.
In-house AI expertise is also in high demand, according to the poll. Half of firms (51%) are concerned that they lack the in-house capabilities to use technology efficiently. Approximately four out of ten respondents (39%) feel lack internal expertise is a barrier to using GenAI.
The survey’s creators highlight the following requirements that successful AI programs must meet:
AI integration: Intelligent decisioning and other decisioning flow tools must be used to easily incorporate GenAI models into decisioning workflows, AI and machine learning applications, and current business processes.
Data protection: Make sure users are secure and private by using robust data quality methods that protect sensitive information, such as encryption, data minimization, anonymization, and synthetic data generation.
Trustworthy and explainable results: Results that are reliable and comprehensible: Data specialists can use natural language processing techniques to preprocess data, reduce hallucinations, and lower token costs. They can also explain the obtained output in simple terms.
Enhanced governance: Use integrated workflows for enhanced governance, which validate all aspects of the life cycle of LLMs, from regulatory compliance to model risk management.
Another duty that needs to be fulfilled is estimating or computing return on investment. According to the survey, over 36% of IT decision makers have either found it difficult to demonstrate that GenAI gives a significant return on investment or anticipate having trouble doing so. Nearly half (47%) of respondents are having trouble applying GenAI in practice after moving from idea to application. A policy on the use of GenAI is absent from four out of ten enterprises (39%).
