AI “hallucinations” to trustworthiness – it is a demystifying journey
Generative AI (read: Gen AI) is not without its constraints.
Perhaps foundation models high-confidence response that is not grounded in the training data, otherwise known as “hallucination” makes their response entirely fictional. While this works for areas like art generation, perhaps even desired as a “creative” feature, hallucinations create invalid responses in areas like legal, healthcare, computer code generation, copywriting, undercutting the potential value of Gen AI.
Looking beyond just input prompts
Generative AI models generate responses based on the dataset it is trained on and the input prompt. Additional sources and datasets cannot be easily integrated into the model’s internal information processing without expensive and expansive retraining or fine-tuning. Absolutely undesirable for enterprises who need a certain accuracy, predictivity as well as contextualization of responses aligned to their own specific data or information.
However, Gen AI models can be combined with other systems using techniques like RAG (Retrieval Augmented Generation) to leverage its benefits. For example, with a chatbot, a conversational AI system can serve as an orchestration layer between the Generative AI model, a search engine, and the user, which helps to amplify the user experience.
Rise of Trustworthy AI
Foundation Models can reproduce “latent bias” in the training data, lacking comprehension and the ability to reason as humans do. This has enterprise-wide implications for the broader concept of Trustworthy AI. After all, they are language models, image models, or voice models, but not knowledge models. Despite limitations, Foundation Models can function at such a high quality that many new use cases become possible.
Trustworthy AI framework for enterprises
Gen AI is a demystifying journey for enterprises. At Mastek, we have designed a 6-pillar, trustworthy AI foundation framework that works across the following aspects to generate accuracy, functionality, and relevance:
1. Stakeholder Engagement
- Chatbots with Gen AI capabilities
- Customer, Partner or Employee self-service
2. Leveraging existing technology investments
- Enable Front-office (CX), Middle-office and Back-office application stack’s Gen AI capabilities
3. Technology productivity gains
- Code generation & Testing using Gen AI
- Full integration of Gen AI throughout the Software Development Life Cycle (SDLC)
- Application Modernization with Gen AI
4. Internal Processes
- Knowledge Management and summarization
- Content creation
- Employee training
5. Customer facing Processes
- Hyper-personalized customer experience with Gen AI
- Reduce knowledge latency
6. Gen AI-based business models
- Full-stack apps built on enterprise or verticalized foundation models
The outcomes are palpable
All these 6 foundation pillars of Mastek’s Trustworthy AI are providing more than satisfactory results across our global clients. Some of these are:
1. Reduction in knowledge latency
2. Acceleration of adoption of digital with AI
3. Minimizing product development time
4. Reimagining business processes
5. Developing Gen AI-powered business models
The road ahead for AI’s new poster child
This new posterchild of AI (read: Gen AI) is stimulating the imagination as enterprises and individuals consider how to use this technology to benefit both business and society.
Trustworthy AI can help in incremental digitization and significant productivity enhancement. But its grander potential is in the higher-order opportunities, such as new services or business models that were previously uneconomical.
That’s where we step in with TRUST, VALUE and VELOCITY.
