Addressing common challenges of Generative AI;

Generative AI like gpt , Gemini is widely used. AI continues to be incorporated with readily available tools suitable for everyday tasks. GenAI does have some issues now which lead to user frustration. Some of the frustrating things which users face with GenAI use include;

• Providing too much information
  • We have been accustomed to using google to search for information. Google results will give you a snippet of information which you can choose to expand for more information. Sometimes a quick result is obtained succinctly which satisfies the questions. GenAI provides a whole lot more information. The user needs to read through this information. Then they must confirm that no information quality issues exist within the output. This is information overload and counter productive to the use of AI in improving efficiencies.
• Including irrelevant sources of information
  • GenAI agents use large language models to pool data from a wide range of sources. These sources may have different uses or interpretations of the same term. Consequently, the information returned can sometimes be irrelevant to the user’s needs. This requires the user to review the information and select the applicable output for them.
• Spelling errors
  • GenAI introduces spelling errors especially when the user is iterating or passing additional information to the agent. Spelling errors in presentations will often distract from the objective and can lead to undesired outcomes for the user
• Applying the wrong context
  • GenAI agents use a wide pool of data. They provide generic answers. These answers are not usually tailored to the user’s needs. Further information needs to be passed on to the agent which allows it to refine its outputs. This can lead to issue highlighted above where spelling errors are then introduced into the output.
• Providing Stale information
  • We live in a world where information changes dynamically. An AI agent is trained on data which is not updated in real time. Therefore, it usually does not provide up to date information.
• Information Bias and Safety Issues
  • These are closely linked to AI context use in generic search

The highlighted issues can be tackled with governance within organisational ecosystem. This will help increase positive adoption of AI which can give incredible benefits when used responsibly.

Governance is often overlooked when organisations try to adopt AI or plunge into AI use without due process. AI promises can only be kept with proper governance. It starts from identifying appropriate data and the collection of data. It also includes AI architecture and audience considerations.

AI’s usefulness can produce positive outcomes. It can also enhance efficiency. However, AI, like most projects or ventures, requires quality data to provide quality outputs.

Look out for vol 2 where I detail the governance guardrails which are effective in overcoming Generative AI challenges.