[Tech World News=Reporter Yang Seung-gap] As big tech companies such as Microsoft (MS) and Google continue to expand investments in AI chatbots, there are concerns about weakening the competitiveness of the AI start-up industry. Related industries have begun to strengthen their competitiveness by providing services that meet the specific business requirements of companies.

Recently, as ChatGPT has received a lot of attention in the market, AI companies have begun to build services, but the problem is securing a business model. In the case of big tech companies such as Microsoft and Google, they can seek to diversify their business structure because they have an advantage in terms of scale and investment, but small start-ups are difficult to change.

According to the related industry, the role of AI such as ChatGPT as a productivity tool is evaluated positively. In the case of MS, it is promoting the application of ChatGPT not only to the search engine ‘Bing’ but also to Office. On the 6th (local time), Google unveiled its interactive AI service ‘Bard’ and is preparing to combine it with the Google search engine.

Currently, as AI service providers in Korea, 'Rutton' of Rutten Technologies and 'Chatbot' of Persona AI are attracting attention. However, large-scale business requirements to compete with big tech companies are burdensome.

Some argue that general-purpose services applied to various industries are important, but to have competitiveness, it is necessary to provide services that fulfill the specific requirements of companies. It is the view that profitability is sought by providing solutions that match the direction of the company.

An industry insider said, “Gigantic AI models such as ChatGPT have characteristics that can be used in various fields.” It is important to respond appropriately,” he said.

He added, "We are working hard to provide optimized options for each company's products and services."

As AI services begin to gain attention as a productivity tool, the importance of data to obtain satisfactory answers has increased. For example, the GPT-3 technology developed by OpenAI is known to utilize 175 billion parameters.

However, the amount of data that is insufficient to learn specific business knowledge compared to the general model is a concern. This is because large amounts of high-quality data are critical to producing accurate results.

An industry insider said, "The number of data required to provide services is not fixed, and the level will vary depending on the purpose and function desired by the company." We have to test it,” he said.

Meanwhile, as generative AI and chatbot services such as ChatGPT draw attention, concerns about potential risks such as hallucination are growing. Because the language model learns from data that has already been stored, it may result in inadequate timeliness or accuracy when receiving information over time.

An industry insider said, “ChatGPT also relies on data learned by 2021, and at first glance it seems to be true, but sometimes it provides incorrect information, so it is difficult to solve the hallucination problem at present.” [Service] providers are constantly looking for combinations of data,” he said.