ANTICIPATE
Determine which model suits your organization Organizations exploring the integration of Gen AI into their operations have three distinct options, each tailored to specific use-case needs, available resources, and desired outcomes.
Use - Case Examples
Talent
Infra
Change
Experiment: Off-the-shelf GenAl applications such as Chat GPT or similar products No confidential data required
• Problem-solving • Code Generation • Idea Generation
Low Medium
Low
1
Integrate: Gen Al applications such as APIs or Open models into existing processes Confidential data required to fine-tune models
• Knowledge Management • Social listening
2
Medium High
High
Build: Your own GenAI models to fit business contexts Confidential data required Train models
• Extensive research and development
3
High
High
High
Requirements Low Medium High
An organization can choose any one approach or combine different approaches based on the use-case need and available talent, budget, and infrastructure
Figure 6. Decision model for Gen AI integration
THE EXPERIMENT APPROACH
This approach involves engaging with ready-to-use Gen AI applications like Chat GPT or similar offerings. This method allows organizations a quick and easy initial foray into Gen AI, without extensive investments or data prerequisites. It offers immediate solutions for problem-solving, code generation, and idea generation tasks. Moreover, the lack of a need for confidential data mitigates potential privacy and security risks. This option is particularly fitting for organizations wanting a risk-free introduction to practical applications of Gen AI.
THE INTEGRATE MODEL
The second option, entails incorporating Gen AI applications such as APIs or open models into pre-existing processes. Though this requires some confidentiality for adjusting models to business-specific contexts, it allows organizations to exploit Gen AI capabilities within their existing framework. While it may require moderate investment and data, this approach allows for greater customization of Gen AI to effectively address business-specific challenges.
© 2023 Fractal Analytics Inc. All rights reserved
07
Made with FlippingBook - PDF hosting