[ad_1]
We at Forrester have been busy in two parallel areas: generating artificial intelligence and top emerging technologies. I’m excited to announce that our new report, The Architect’s Guide to Generative AI, is where these two streams come together. Generative artificial intelligence (genAI) is our most important emerging technology in 2023, and will likely remain so in 2024. For our top 10 technologies, we are working on three reports: “Status,” “Future,” and “Architect’s Guide.” Since the summer, we’ve been rolling out “what’s here” and “what’s next” reports on top emerging technologies. We just released our first architect guide covering genAI.
The Architect’s Guide report is designed to provide architects of all types with a deeper understanding of top emerging technologies and provide insights into the security and risk aspects of that technology. It follows a very easy-to-recognize architecture process: we discuss current and future trends, break emerging technologies into functional building blocks, assemble the building blocks into solution patterns, and then architect them with technology as well as governance, security, and risk considerations.
GenAI Architecture goes well beyond the scope of the LL.M.
Architects are caught up in an extreme passion for execution and the need for speed. Demonstrations of GenAI through ChatGPT and other means look amazing and have executives excited about the potential. However, behind the scenes, we found that implementation quickly became challenging. Architects are in a dilemma – should you wait for genAI to appear in the software you already use, or build something from OpenAI or other software? Or will you venture into the open source world with models like Llama and Mistral? we discover:
- Generative AI goes far beyond a single LL.M. We identified four ways in which genAI can enter organizations. We also looked at who is moving beyond “bringing AI” or simply using genAI tools within existing software. As you can imagine, most companies are talking about using upcoming genAI-infused enterprise software or experimenting with OpenAI, Microsoft, Google, and other public model-as-a-service providers. At the most advanced level, most of the time and effort spent has little to do with the primary LLM chosen.
- A RAG-centric solution architecture with pipelines, gates, and service layers works best. Incorporate Retrieval Augmentation Generation (RAG) into your solution architecture. Generating intent and governance gates at both ends of the model is critical. GenAI solutions are being rolled out through pipelines focused on rapid shaping of the front end and governance for outcomes. We are seeing an increasing need to pay more attention to the front end, intent recognition, prompt processing and model grounding through RAG. Most companies trying to fine-tune model parameters tell us that instant shaping is just as effective as base and is cheaper and faster, at least initially.
Don’t ignore the fundamentals
While genAI architecture may seem like a moving target, don’t lose sight of the fundamentals. Your technology architecture must be flexible to handle more demands—more models, more data, and more scale seem to be the order of the day. To be successful, your technology stack must be able to manage and version more models, monitor more results and more engagements, and optimize costs as everything gets bigger. Furthermore, this doesn’t just exist in the cloud; You need to prepare for personal artificial intelligence devices to be in your pocket soon! focus on:
- Collect well-managed data. Whether structured or unstructured, consistent, well-stocked, and high-quality data is critical if your goal is to apply genAI to your own organization’s problems.
- Address governance, risk and compliance stakeholder concerns. In the Architect’s Guide, we embrace the idea of building specific architectural models that address the concerns of risk-oriented stakeholders. In our report we provide you with examples of how to resolve these issues.
- Participate in governance discussions based on a set of principles. Don’t try to establish standards; things move too fast. In the report, we identify three data and application principles and three infrastructure and operations principles to guide your efforts.
Many thanks to my colleagues Charlie Betz, Jeff Pollard, Alvin Nguyen, and Will McKeon-White for accompanying me on this journey of discovery as we delved deeper into this technology, and to several other analysts who shared their experience of. Take a closer look at this technology. Thanks to several architects and technology leaders who volunteered their time to participate in this research, and special thanks to our internal genAI team for their work on Izola, our customer-facing research tool is in beta and will be rolled out to all our customers soon. .
[ad_2]
Source link