February 15, 2022 Current Natural Language Processing AI

Tetsuji Yamashita
Corporate Director (CXO) / Head of Division Four / Manager, Digital Transformation Dept.

I'm always looking for ways to increase productivity by integrating digital technologies such as AI and RPA into business processes, and to create services that are unique to our company. While there are a number of digital technologies that I'm interested in, I'm particularly drawn to the field of natural language processing. In particular, OpenAI's GPT-3 (Generative Pretrained Transformer 3) text generating large language model, initially released in July 2020. This model is highly regarded and has a wide range of applications, including writing stories, press releases, and programming code, and for design work. Overseas, several commercial services have already been established.

Now, in this Column, I'm going to share the results of our real-world tests of the GPT-3's capabilities and talk about where we are regarding natural language processing. First, please read the text we created using GPT-3 on the topic "RPA trends in 2022".

RPA Trends in 2022

What is RPA? How Does it Work?

Robots and automation are becoming increasingly common in the workplace. One of the most popular forms of automation is Robotic Process Automation, or RPA.

RPA is used around the world for a wide variety of tasks - from invoicing and data entry to document management. RPA has been proven to save companies money and time and provide greater accuracy than if humans were to perform the same tasks.

Use of RPA in Different Industries

The use of this technology is not limited to any particular industry or sector. Companies across all industries are looking for ways to automate their processes and reduce costs.

Some of the industries in which RPA is being used include -

Banking
Healthcare
Manufacturing
Retail

What are the Advantages and Disadvantages of RPA Implementation?

Advantages

Reducing labor costs by automating manual tasks.
Improving efficiency by eliminating duplication and simplifying processes.
By outsourcing less important tasks to robots, companies can concentrate on their core competences.

Disadvantages

As sensitive data is increasingly transferred electronically, between computers and robots, without appropriate security measures there are increased cyber risks.

The Future of RPA

RPA continues to evolve. In the future, for RPA in industry, workflow automation and the creation of machine learning models using artificial intelligence will be important.

How do you feel about this? You only need to enter the theme of the text you want to write and a few keywords, and the model will generate a content in less than 30 seconds. Then after a few minutes of editing, including unifying terms and removing overly verbose parts, you'll have a text of the above quality. I believe that currently the quality is fairly good and it's possible to produce something that resembles professional prose in such a short time.

However, if you have some background knowledge in RPA and are reading this Column, you may find that the article resembles advances that you have seen or heard elsewhere, and that the article is unworthy of the exaggerated title of "RPA trends in 2022".

Well, this is just one example, and I chose a difficult subject for AI - that is predicting the future. This Column highlights what current language AI models can do and their limitations. We're still a long way from the fantasy world where AI can output more creative results than humans, or even replace them. However, depending on how it is integrated into work systems, AI can already be used to a certain extent for creating quick, draft-level documents for blog posts, or for summarizing known facts.

I used an overseas service to create this text. Nowadays, services based on language AI models such as GTP-3 are continuously being released globally, and they are evolving at a startling pace. While being alert to latest innovations in this space, we will continue to gather information and implement tests to find ways to best apply such models to our services and business processes.

If you’re interested in automatic text generation and multilingual translation using AI, or if you are interested in the future of the web production process, please do not hesitate to contact us.