Got a question about large language models? Three out of four app developers might say: Ask ChatGPT.
An IBM-sponsored survey, released last week, revealed that less than one-quarter of enterprise application developers consider themselves generative AI “experts.”
According to the report, the small pool of confident GenAI pros demonstrates a “skills gap” and “steep learning curve” for IT pros navigating a space with evolving toolkits and frameworks.
“Any simplified AI application that can simplify the workload and enhance productivity? Creating that is actually not simple, and it’s complex and it’s very dynamic in a world that is changing rapidly every day,” Maryam Ashoori, senior director of product management for watsonx models at IBM, told us.
The survey, conducted by Morning Consult, polled more than 1,000 US-based developers, engineers, and data scientists building generative AI applications.
Some report findings:
- Only 24% of application developers surveyed ranked themselves as GenAI experts.
- Top challenges included “lack of a standardized AI development process” and “developing an ethical and trusted AI life cycle that ensures transparency and traceability of data.”
- Almost all (99%) developers are exploring or developing AI agents, what IBM defines as “a system or program that is capable of autonomously performing tasks on behalf of a user or another system by designing its workflow and utilizing available tools.”
Deloitte’s 2024 Q3 study of 2,770 director- to C-suite-level respondents found two-thirds of orgs said they are increasing GenAI investments “because they have seen strong early value to date.” (On the confidence point, however, only 23% rated their organization as highly prepared.)
A July 2024 Gartner report cited real-world GenAI examples, including prompt tools to assist with vehicle-design sketches; call-center assistance for banks; and transportation data analysis to determine disruption causes.
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
And now AI agents have arrived, requiring generative functions to perform a specific task; a chatbot “agent” may generate human-like text. An enterprise scheduling assistant may generate time-slot suggestions.
Agent development calls for knowledgeable choices of large language models and orchestration frameworks, Ashoori said.
“Are you going to accept the pre-built agent that is coming to you as a black box? To just give access control to your data sets for these agents to, for example, interpret code, run something on your data sets,” Ashoori told us, noting the importance for developers to understand data traceability.
Companies like IBM have tools like the chat-based watsonx Code Assistant to address a lack of expertise with AI application development tools. Salesforce, in December 2024, introduced a 2.0 version of its Agentforce platform, which allows users to build customized agents using natural language.
“I think for a lot of engineers working in this space, it is hard to stay on top of all of the new breakthroughs, all of the new technologies, everything coming out from every company. So, it’s hard to say you’re an expert in a space that is moving this quickly,” Alice Steinglass, EVP and GM of Salesforce’s AI platform, told IT Brew. Steinglass is particularly excited about 99% of developers experimenting with agents.
“The important question is not where we are today, but are we learning it? Where are we going with it?” she told us.