GenAI Tools Used by 88% of US Businesses, What Problems Need Solutions?

Generative AI tools have caused a revolution in the tech and business sectors, with ChatGPT, Midjourney, Google Bard, and Bing Chat leading the way. A recent study by TECHnalysis Research revealed that an impressive 88% of US businesses have already embraced GenAI tools, but there are still challenges to overcome. The study’s findings shed light on the emergence of “prompt engineers” as a new job category and the various concerns surrounding GenAI adoption.

The demand for integrated prompt training

Fig. 1 illustrates that 62% of respondents strongly desire integrated prompt training in future GenAI tools. Prompts play a vital role in maximizing the effectiveness of these tools, making it essential for vendors to respond to this request. While “prompt engineers” have been touted as a new job category, democratizing the ability to create effective prompts is crucial to mainstream adoption. Vendors must ensure all users can obtain the best results from GenAI tools.

Customized reinforcement learning and Its complexity

Customized reinforcement learning, the second most requested feature, reflects the desire for personalized foundation models that integrate human feedback, like Reinforcement Learning from Human Feedback (RLHF). 6 out of 10 survey takers expressed interest in this capability, demonstrating a level of sophistication among respondents. However, the study also revealed potential confusion or misunderstanding about this feature. The desire to customize GenAI models in various ways remains high on the priority list for most respondents.

The need for watermarking and packaging clarity

A striking 91% of IT decision-makers surveyed indicated a preference for watermarking outputs from GenAI tools. However, the lack of automatic watermarking mechanisms in current GenAI tools presents a clear opportunity for vendors to improve their offerings.

The packaging and selling of GenAI models also present challenges, with a wide range of opinions and confusion among respondents. Many find it difficult to distinguish whether vendors offer standalone GenAI models, applications, or services powered by GenAI or other variations. Vendors need to clarify their offerings and integration possibilities to aid customers in making informed decisions.

Diverse approaches to bringing GenAI Tools to market

Figure 2 reveals a diverse set of preferences among respondents regarding how GenAI tools should be brought to market. Over 51% believe integrating GenAI capabilities into existing applications is the best approach. Around a quarter prefer standalone models, while the rest lean towards an as-a-service model for specific or any applications. Vendors must experiment with various methods to cater to customer preferences effectively.

Respondents were also divided on whether core foundation models should be integrated into each app or shared across multiple apps. Approximately 55% advocated for separate foundation models for each app, while 35% favored a single “master” model shared across multiple applications. The remaining 10% suggested multiple models shared across various apps. Balancing these preferences poses a challenge for GenAI vendors.

Educational and marketing challenge for GenAI vendors

The study highlights the considerable educational and marketing challenge faced by GenAI vendors. Articulating the capabilities of their tools, compatibility with other systems, and seamless integration into existing workflows is crucial to empowering customers to understand their options and achieve their goals effectively.

The excitement surrounding GenAI tools is palpable, and they have already found applications across various industries. However, challenges remain, and the study illustrates the need for ongoing development and education. As GenAI tools continue to evolve, vendors must work diligently to clarify their offerings, and customers must be clear about their requirements to make the most of these revolutionary tools. The future holds promise for GenAI adoption, but it will be an interesting journey of learning and understanding for all stakeholders involved.