Generative AI
Generative AI refers to deep-learning models that can generate high-quality text, images, and other content based on the data they were trained on.
Generative AI is unlike any technology that has come before. It’s swiftly disrupting business and society, forcing leaders to rethink their assumptions, plans, and strategies in real time.
To help CEOs stay on top of the fast-shifting changes, the IBM Institute for Business Value (IBM IBV) is releasing a series of targeted, research-backed guides to generative AI on topics from data security to tech investment strategy to customer experience.
- Executives struggle to assemble the right mix of AI models to drive innovation while managing costs and risks.
- Average company is using 11 generative AI models today, 17 in 3 years.
- Considerations for CEOs: model type, size, and ownership.
- Commercial models like GPT-4 only represents 25% of models used at typical organizations.
- 63% of executives citing model cost as a top concern.
How GenAI impacts the development of pipelines?
🔹 Boilerplate generation is a huge win here that allows us to focus on the business details that matter! I cover a lot of this in more detail here
🔹 Implementing many quality checks and write-audit-publish can be done with a single prompt nowadays.
How GenAI impacts the maintenance of pipelines?
🔹 Troubleshooting false positives and out-of-memory errors will soon be a thing of the past!
How GenAI will move data engineers to two directions?
🔹 Towards the business and analytics. I call this the merging of the data analyst and data engineer.
🔹 Towards the server and online application. I call this the merging of the data engineer and the software engineer.
The Ultimate Walkthrough of the Generative AI Landscape
Generative AI and LLMs are fast becoming a game-changer in the business world. Everyone wants to learn more about it.
There’s a gen AI model for that
ChatGPT made everyone feel like an AI expert. But its simplicity is deceptive. It masks the complexity of the generative AI landscape that CEOs must consider when building their AI model portfolio.
Generative AI models come in many flavors. What they can do, how well they work—and how much they cost—varies widely. Who owns the model, how it was developed, and the size of its training dataset are just a few of the variables that influence when and how different models should be used.
With the massive amount of data and resources it takes to train a single large language model (LLM), the question of size is monopolizing many conversations about gen AI. As a result, many CEOs wonder whether they should scale large gen AI models for their business. Or if they should develop smaller, niche models for specific purposes.
The answer is, they need to do both. And many already are.
A typical organization uses 11 generative AI models today—and expects to grow its model portfolio by ~50% within three years.
Why so many? Because every use case comes with its own requirements and constraints. And different business problems demand different types of models.
For example, tasks that are highly specialized, such as image editing or data analysis, need gen AI models that are trained on smaller, niche datasets. Work that is sensitive or proprietary requires gen AI models that can be kept confidential— and close to the vest. More general tasks, such as text generation, may call for gen AI models trained on the largest datasets possible.
While CEOs should have teams that understand all the details about what sets different models apart, you do need to know that picking the right model for each task—each application of generative AI—matters. Knowing what drives cost, environmental impact, and business value will help you optimize the performance of your AI portfolio—and give your teams the tools they need to beat the competition.
Tech Stack
GenAI refers to systems capable of creating new content, such as text, images, code, or music, by learning patterns from existing data. Here are the key building blocks for GenAI Tech Stack:
- Cloud Hosting & Inference: Providers like AWS, GCP, Azure, and Nvidia offer the infrastructure to run and scale AI workloads.
- Foundational Models: Core LLMs (such as GPT, Claude, Mistral, Llama, Gemini, Deepseek) trained on massive data, form the base for all GenAI applications.
- Frameworks: Tools like LangChain, PyTorch, and Hugging Face help build, deploy, and integrate models into apps.
- Databases and Orchestration: Vector DBs (such as Pinecone, Weaviate), orchestration tools (such as LangChain, LlamaIndex) manage memory, retrieval, and logic flow.
- Fine-Tuning: Platforms like Weights & Biases, OctoML, and Hugging Face enable training models for specific tasks or domains.
- Embeddings and Labeling: Services like Cohere, Scale AI, Nomic, and JinaAI help generate and label vector representations to power search and RAG systems.
- Synthetic Data: Tools like Gretel, Tonic AI, and Mostly AI create artificial datasets to enhance training.
- Model Supervision: Monitor model performance, bias, and behavior. Tools such as Fiddler, Helicone, and WhyLabs help.
- Model Safety: Helps ensure ethical, secure, and safe deployment of GenAI systems. Solutions like LLM Guard, Arthur AI, and Garak help with this.
Videos
Generative AI frameworks
