[Correspondent Mecl Goes🏃] Generative AI Strategy for Business Leaders: Finding Centering Amid Skepticism and Hype
Date of creation : 2024-05-22
Hello everyone! The Mekl correspondent is back with the third AWS Summit session review. 🙌 Thanks to the overarching theme of 'Generative AI' throughout the conference, I was able to encounter a lot of related content not only in the keynote but also in various sessions. Today, I would like to summarize two sessions that dealt with <Generative AI Practical Implementation Strategies>.
As the complexity surrounding generative AI increases with skepticism and hype, I believe this will be very helpful for business leaders who are pondering how to maintain their focus.
The sessions I attended were…
📑 Practical methodologies for applying generative AI to enterprise business | Seo Gil-joo, Group Leader, Megazone Cloud AI & Data Center
📑 Six generative AI strategies for leaders: Learning from the past and designing the future | AWS Kang In-ho SA, Eo Han-na SA
With the rapid increase in data, the availability of scalable computing power, and innovations in machine learning, generative AI is now at a significant turning point. Many people compare this to the invention of the light bulb in the 19th century and the emergence of the internet in the 20th century as a massive technological innovation. However, the pace of technological innovation is accelerating, and there is always confusion during such transitional periods.
Seo Gil-joo, Group Leader of Megazone Cloud's ADC, states that technological innovation at a cultural level has both side effects and positive effects, and there was similar confusion when the technology of cloud computing first emerged about ten years ago. This is because the way people work has completely changed, from business managers and product managers to developers and operators. Specifically, he pointed out issues such as weakened governance control due to 'shadow IT', the occurrence of technical debt due to a lack of standardization, and information leaks.
However, as mentioned earlier, such technological innovations have always existed in human history, and we can learn lessons on how to respond when new technologies emerge.
The following are six lessons learned from history presented in the <Six Generative AI Strategies for Leaders> session, and I would like to highlight a few particularly noteworthy points.
Successful people used the right tools.
A solid foundation was established beforehand.
There will be skepticism and hype.
Reimagine work with new technologies.
Change brings new risks and responsibilities.
The value of new ideas grows exponentially.
What does it mean to use the right tools in the era of generative AI? It means choosing the methodology that is most suitable for our business among various generative AI approaches. As a criterion for such choices, Megazone Cloud presents the 'Generative AI Promotion Framework'.
Which processes are generative AI applied to?
What data does the process utilize?
What are the detailed flow and functional/non-functional requirements?
What is the optimal technology to respond to this?
The key here is well-organized data that fits the process, an efficiently referenced data catalog, and finally, an interface that can utilize various generative AI models.
An interface that can utilize various generative AI models is Amazon Bedrock. AWS Kang In-ho SA explained the strengths of easily applying various foundation models through a unified API and the ability to ensure safe data protection through Guardrails for Amazon Bedrock.
Before introducing generative AI, there are foundational elements that must be established. These are a cloud that can flexibly scale and well-refined data. The specific methods can be found in <Six Generative AI Strategies for Leaders>.
Methods to Accelerate Cloud Journey
Establish cloud adoption goals/principles → Establish cloud-based elements → Training → Practice/experience → Workload migration → Utilize experienced partners
Methods for Data Refinement
Collect and expand unbiased diverse datasets → Focus more on data quality → Manage data history/change tracking → Version control of data used in models → Improve speed through automation → Cost awareness and management
Additionally, AWS Kang In-ho SA shared the following statistics and discussed what role leaders should play amidst skepticism and hype:
The percentage of jobs in the US and Europe exposed to some AI automation is 66%
The percentage of employees needing retraining within the next 5 years is 50%
The percentage of employees working in jobs that did not exist in 1940 is 60%
The point is that while looking at the same statistics, some people may have fears about change, while others may find motivation. Therefore, it is important for leaders to understand the various perspectives brought about by uncertainty and to gain full agreement and participation from the team.
The final presentation introduced the 'Generative AI Project Lifecycle' for the explosion of value in new ideas. The key is to select suitable projects that can pursue business value and to repeat testing and evaluation.
Selection: Choose existing models or create models from scratch
Scope: Define use cases
Model adjustment and alignment: Prompt engineering, fine-tuning, incorporating human feedback
Application integration: Deployment for model optimization and inference, building applications utilizing LLM
Now, I have summarized two sessions of <Generative AI Practical Implementation Strategy> that would be helpful for business leaders. I hope this has been beneficial for you as well. Megazone Cloud provides end-to-end services, from consulting on target business areas to construction and operation, so that companies can utilize immediate generative AI based on their unique data. If you are curious about Megazone Cloud's ‘GenAI360’ service, please visit the site.
Check out the keynote reviews too!
👉[Mekel Correspondent Reports🏃♀️] Generative AI Strategy Confirmed at AWS Summit 2024 Keynote
👉[Mekel Correspondent Reports🏃♀️] Three Cloud Technology Trends Every Engineer Should Know
