Companies that wait to adopt AI might never catch up.
Most companies are fast followers because they ‘wait and see’ what others are doing before transforming.
This fast follower strategy will not work for AI.
There are three key reasons for this:
1) Choosing and implement the right AI stack based on the needs of the organization around internal and external use cases, data privacy, and security. The architecture decisions become challenging given how dynamic this space is. There are AI innovations and accleration happning almost every week.
2) Expert human need to spend enough time with AI to provide direction, nudges and learning. Even if we use RLHF, there is significant effort required to provide the rewards-based learning and train AI. This would apply to either a broad LLM like OpenAI‘s GPT 4, Google‘s PALM 2; or a narrow LLM like what Intuit and Bloomberg are creating. Also, the training needs be aligned to responsible AI principles.
3) Ongoing AI governance against model drift: AI applications need a comprehensive governance approach. AI algorithms, built on historical data, decay over time and need continuous monitoring for potential bias and interpretation of changes in business context.
If companies don’t start early, it is hard to build muscle over a period of time.
If your company is not adopting AI, your competition likely is. With AI, the competition can disrupt incumbents not in decades, but in years or even months.
#ai #aigovernance #aistrategy #llms #llm
Further reading, though slightly dated: https://lnkd.in/gB54xX8F
