New
Gear Up Your AI: Fine-Tuning LLMs
December 29, 2023

Predictions for 2024

Table of Contents:

 

Predictions 2024

My Predictions for 2024

What a wild and thrilling ride we had in 2023! Clarifai celebrated its tenth anniversary by launching the world's first full stack AI Platform. We continued to lead the way forward through the generative wave, bringing speed, efficiency, and simplicity to developers building generative AI for themselves and the organizations they support. 

So now, my predictions for 2024:  

 

1. Generative AI hype will slow by mid to late 2024 as the market focuses on ROI

With market projections in both computer vision and genAI growing exponentially, the hype surrounding AI is extraordinary. However, to justify the market investments and growth, the hype will slow mid to end of 2024 as the focus shifts to actual ROI. Organizations will need to get beyond prototypes and deliver positive and sustained ROI from their investments.

2. A hybrid cloud strategy will be preferred for enterprises adopting AI that is consistent with their existing infrastructure

As enterprises start to understand and focus on real ROI from their AI investments, the environment that the AI runs in will also become a central focus. The infrastructure cost to run these huge models is significant, especially when compared with a global GPU shortage, so more companies will look to hybrid cloud as a preferred alternative, where they can leverage their existing infrastructure for compute-intensive work rather than paying cloud providers.

3. A shift from LLMs to fine tuned task-specific models for accuracy, cost, privacy, etc., will occur 

People are starting to understand that large LLMs are great for general-purpose use cases, given their ease of use and flexibility. When you think about a specific business problem you are trying to solve, you don't want or need a general-purpose solution. I expect to see a significant shift to fine-tuning models for specific tasks that deliver better accuracy, cost less, and provide better privacy because you can control precisely what data they're trained on.

4. Vendor consolidation towards platforms rather than tools like vector DBs, embeddings, LLMs, open source, etc., pieced together separately

Following the 2023 growth of new entrants and AI tools, there will be a natural consolidation as we move forward. Just look at AI vector databases as an example. Vector databases store a representation of your data, something we've had embedded in our platform since 2016. Today, there are about 50 different vector database providers, and that's all they provide. The market cannot continue to sustain 50 of the same kind of offerings,  which is why the shift to consolidate into platforms will emerge. Platforms reduce the number of unique tool decisions and integrations organizations need to take on not just for the initial purchase but for ongoing support, maintenance, security, and privacy considerations - not to mention keeping up with the rapid changes in state-of-the-art technology. 

Conclusion 

2024 will be the most important year ever for AI, encompassing new technology developments, business impact, regulatory policies, and ethical and social implications.  Enterprises will need to start proving ROI from both everyday and transformative AI investments. Getting it right will have an enormous impact on value generation across businesses and the world.

Wishing everyone a healthy, prosperous, and Happy New Year! Thank you to our loyal clients - developers, data scientists, engineers, business professionals, analysts, and anyone else - who have been on our journey to date. We are eagerly looking forward to continuing to bring an array of exciting offerings your way in 2024.  If you haven't tried Clarifai yet, today's a great day to start for free.