Discussions about benefits and dangers with artificial intelligence is a global focus. A new survey shows 70% of business leaders believe that artificial intelligence is either critical or highly critical to their business in the next three years. And 72% are increasing investments in AI in 2023, across nearly every industry, according to a survey of 1 600 machine learning practitioners and business leaders, Vijay Karunamurthy, field CTO at San Francisco-based startup Scale AI, writes in a blog post for World Economic Forum.
60% in the survey say they are experimenting or plan on working with generative AI in the next year but only 21% have models in production. The survey shows 59% of companies view AI as critical or highly critical to their business in the next year, and 69% in the next three years.
“This reflects the amount of work that it actually takes to make generative AI work for enterprise use cases. These models are not just plug-and-play. Domain-specific business data needs to be carefully fed into these models using a process called fine-tuning, and the models need to be aligned with company policies and desired behaviour using a technique called reinforcement learning from human feedback (RLHF).”
For most, there is a skills and technology gap to overcome to make generative AI enterprise-ready. One-third of respondents cite a need for more expertise, software and tools as their most significant challenge in implementing generative AI, Karunamurthy reports.
“Since most companies don’t have the necessary resources or mandate to build generative models from scratch, they must rely on third parties.”
“Of those companies that plan on working with generative models, the vast majority are looking to leverage open-source models (41%) or Cloud API models (37%), while very few are looking to build their own (22%)”, Karunamurthy writes.
“There are exceptions, but these are not the norm. Training models from scratch is expensive, time-consuming and requires deep machine learning expertise. But, by using existing base models that are fine-tuned with proprietary company data, you can get similar performance without building your own model.”
“When properly adopted, the impact of AI is overwhelmingly positive, with 89% of respondents benefitting from the ability to develop new products or services, 78% from improved customer experience, and 78% from enhanced functionality of existing products or services.”
Scale AI predicts the following trends over the next year:
- Increased investment in AI: As generative AI is now more capable and widely available, companies are quickly incorporating it into their operations. 72% of companies will significantly increase their investment in AI each year for the next three years.
- Growth in application of fine-tuned models with proprietary data: On their own, base generative models are valuable tools. Paired with a business’s proprietary data, they become strong differentiators, improving customer experience, product development and profitability.
- Models will also become multimodal: Models will be able to consume and generate text, images and video, making them even more useful than they are today. Capable assistants can extend developer, data science and business analyst productivity.