Get Started with Generative AI on AWS

Introduction

Large-Language Models (LLMs) and multi-modal models are revolutionising tasks like code generation and image creation from text. These models, termed foundation models (FMs), are customizable for specific needs without starting from scratch each time. Pre-training involves training on vast amounts of unlabeled data, resulting in impressive performance across various tasks. This ebook offers an overview of FM capabilities, discusses opportunities and risks, and demonstrates leveraging LLMs with AWS technologies for secure solutions.

Selecting and customising foundational models

Zero-shot learning allows non-ML experts to interact with Foundation Models (FMs) through web playgrounds or chat interfaces like ChatGPT. By providing natural language commands (prompts), users can perform tasks such as listing action items from meeting transcripts or translating documents.

In-context learning enables developers to improve model outputs by including examples within input prompts. For instance, a prompt to create a social media ad for a product can be enhanced with examples of past ads for similar products.

Fine-tuning allows customization of FMs for specific tasks using a small number of labeled examples. This approach is cost-effective and efficient, requiring far less labeled data than building a task-specific model from scratch. For example, a recruiting firm can fine-tune an FM to automatically process resumes and generate summaries at scale with just a few examples.

---

Issues to consider with foundational models

Responsible AI
Foundation models introduce new challenges in ensuring responsible AI throughout the development cycle, including issues of accuracy, fairness, intellectual property, toxicity, and privacy. These challenges arise from the vast size and open-ended nature of foundation models compared to traditional machine learning approaches.

Hallucination
Another concern is hallucination, where Large-Language Models (LLMs) generate inaccurate responses inconsistent with training data due to the way they represent inputs.

High Costs
Another issue is the high costs associated with real-time inference generation, prompting the need for smaller, more fine-tuned models for specific use cases.

Emerging concerns with foundation models include toxicity, where generated content may be offensive or inappropriate, and intellectual property considerations, as LLMs occasionally produce verbatim passages from training data. Despite these challenges, ongoing efforts such as user education, content filtering, and technical solutions like watermarking and differential privacy are being developed to address them.

---

The opportunities ahead with generative AI

Generative AI has the potential to bring about sweeping changes to the global economy. According to Goldman Sachs, generative AI could drive a 7% (or almost $7 trillion) increase in global GDP and lift productivity growth by 1.5 percentage points over a 10-year period. Much of this growth is driven by spend on generative AI cloud services which are estimated by Bloomberg to reach over $109B by 2030, a CAGR of 34.6% from 2022 to 2030.

Sources: Generative AI could raise global GDP by 7%, Goldman Sachs, April 2023 Generative AI market to be worth $109,37 billion by 2030, Bloomberg, January 2023

---

Build with generative AI on AWS

Innovate with generative AI
With enterprise grade security and privacy, a choice of leading foundation models, a data first approach, and the most performant, low-cost infrastructure, organisations trust AWS to deliver generative AI fueled innovation at every layer of the technology stack.

Securely build and scale generative AI applications
AWS also offers the most comprehensive set of services, tooling, and expertise to help you protect your data, so it remains secure and private when you customise and fine tune foundation models.

The most performant, low-cost infrastructure
Train your own models and run interference at scale. With AWS, you get the most performant and low-cost infrastructure for generative AI and the broadest choice of accelerators in the cloud.

Data as your differentiator
With AWS, it’s easy to use your organisation’s data as a strategic asset to customise foundation models and build more differentiated experiences.

---

About Devoteam

Devoteam’s approach is comprehensive, addressing the needs of both enterprises and end-users. We believe in anticipating challenges and proactively finding solutions. With Devoteam, you’re not just navigating the AWS cloud – you’re embarking on a transformative journey with an award- winning partner by your side. With a rich history, a passionate team of experts, and a track record of success, we are here to help you harness the full potential of AWS.

---

Benefits of building generative AI solutions on AWS with Devoteam

At Devoteam, we understand that harnessing the potential of Generative AI is crucial for modern businesses. We are experts in not just advising on Generative AI, but actively implementing it into production for our Enterprise customers, unlocking new business value opportunities and igniting a wave of innovation. Our AWS Generative AI practice brings the best AI and data experts together, to deliver AI Transformation Consulting, AI Solution Development, and AI Capability building.

---

Case Study: LightOn

See how Devoteam migrated LightOn to the AWS Cloud an enabled them to be a pioneer in Gen AI

Challenge
In order to support its growth and accelerate the production of ML models and their availability, LightOn chose to migrate to the AWS Cloud, and in particular to rely on Amazon Sagemaker. The teams from devoteam supported LightOn in this project, with the objectives of:
– Creating an AWS Landing Zone using Terraform, and in compliance with AWS best practices.
– Migrating and optimising resources from another cloud hyperscaler to AWS. Automating the process of creating Machine – Learning models, and making them available on the AWS Marketplace.

Solution
Devoteam and LightOn streamlined and accelerated LightOn’s processes by utilising Amazon SageMaker to automate ML model production and their availability on the AWS marketplace, and the provision of infrastructure on demand while ensuring scalability, stability, and cost effectiveness of powerful GPUs.

Results
The work provided on this project was of high quality, with great expertise on both the management of the migration and the deployment of the solutions to the AWS Marketplace. LightOn now have infrastructure deployed with Terraform, and ready to use, to support their Gen AI platform.

---

By downloading our ebook, you’ll gain insights into:

• The transformative power of generative AI
• Success stories of leading companies leveraging AWS ML capabilities
• Devoteam’s expertise in implementing generative AI solutions for enterprises
• Strategies for harnessing the potential of Generative AI in your business