laitimes

IBM experts' view: Enterprises need to dispel the myth of AI applications, and "talking to data" will reshape the business model

BEIJING, July 26, 2024 /PRNewswire/ -- Recently, the World Internet of Things 500 Summit was successfully held in Beijing. Focusing on the theme of "Smart Connecting the World, Win-win Future", the summit held the opening ceremony, the Fortune 500 Enterprise Forum, the International Cooperation Forum, and the release of the 2024 World Internet of Things Top 500 Rankings. Liu Shengli, Technical Director of Data and Artificial Intelligence of IBM Greater China, was invited to attend the meeting and discussed with enterprise representatives in the "Fortune 500 Enterprise Forum" on new views on the digital economy of the Internet of Things and the relationship between the Internet of Things and the digital economy.

IBM experts' view: Enterprises need to dispel the myth of AI applications, and "talking to data" will reshape the business model

Liu Shengli, Technical Director of Data and Artificial Intelligence for IBM Greater China

Liu Shengli said that in the Internet of Things and even a wider range of application scenarios, enterprises have been actively implementing artificial intelligence technology, but there are "myths" such as too few models, bigger the better, and out-of-the-box use. Taking AI assistants as an example, the interaction form of enterprises at this stage is still dominated by "dialogue with applications", and whether enterprises have the ability to "directly talk to data" in the future will become the key to reducing costs, increasing efficiency and improving competitiveness. He believes that no matter how "volatile" the current industry competition is, we should believe that science and technology can create a better life and future for mankind.

In addition, IBM was selected into the "2024 World Internet of Things Top 500 List" released by the conference, and ranked 11th in the "Top 100 Gold List" list.

IBM experts' view: Enterprises need to dispel the myth of AI applications, and "talking to data" will reshape the business model

IBM experts' view: Enterprises need to dispel the myth of AI applications, and "talking to data" will reshape the business model

Here are excerpts from his remarks (some of which have been reedited):

Three myths about AI applications: too few models, bigger is better, and out-of-the-box use

IBM is a veteran IT company with a history of more than 113 years, and before the 90s of the last century, the history of IBM's development can be said to be the history of computer development. Today, IBM is a technology company focused on hybrid cloud and artificial intelligence (AI), and we are very focused on bringing these two innovations to life.

In the past two years, the discussion about generative AI and large language models has been very hot around the world, and everyone doesn't seem to see IBM particularly "rolling" on this track. In fact, we have done a lot of solid work around the world, especially in the implementation of large models in enterprises.

This also leads to the first observation I want to share today: enterprises do not need "volume" large models, and the current large models are not insufficient, but too many; A lot of time is actually invested in choosing a model. Taking the HuggingFace open source platform as an example, it already has more than 700,000 open source models, which is currently enough.

Second, everyone is concerned about the ultra-large language models that support ChatGPT, but for most companies, the model parameters used are usually between 1B and 70B, which is enough to ensure that the cost is controllable and the performance is up to standard. Very large models are unnecessary and very costly.

Third, when an enterprise chooses a large model, it will find that the matching degree with its own business is often less than 60 points (out of 100 points), and in some scenarios, it cannot even reach 30 points. IBM's large number of practices and implementation cases in the global market, including the Chinese market, are helping enterprises to choose good models and scenarios, and increase the score from 30 to 80 to 100.

"Talking to Data" will reshape the way businesses operate

In the Chinese market, IBM has served a large number of intelligent manufacturing enterprises in the past seven or eight years, including leading customers in automobile manufacturing, new energy and other industries. Based on my experience with some of our clients' top executives, I would like to share two perspectives:

First, today's advanced manufacturing companies are still talking about concepts such as "unmanned factories", but in the near future, we will usher in "

Unmanaged

Or a "less managed" business. The so-called "unmanaged" does not mean that people are not needed, but that people are freed from repetitive, transactional work and can be focused on more creative work. Taking IBM as an example, we are already using AI technologies such as watsonx to accelerate intelligence and automation in different business areas such as HR, IT, supply chain, and internal operations.

Second, "

Talk to your data

It will help enterprises achieve "unmanned management". When I recently communicated with a customer of a domestic car company, we imagined a scenario of "talking to the data" in the traditional sense, rather than "talking to the application": a sales leader asked "the sales trend of model A in Beijing in the past three days", "why did such a change occur", and "what actions should the company take" based on the sales data of the past few days. in other words

Not only can you find out what happened, but you can also find out why it happened and plan your next steps.

In the past, these jobs relied on large numbers of white-collar "knowledge workers." Today, this knowledge is being learned quickly by AI and done in a more accurate and efficient way. For example, with the newly released IBM watsonx BI Assistant, we can talk directly to the data without having to write a program. Constant questions from professionals can help AI continue to learn and become more professional. Professionals can also improve themselves in the process of dialogue and master more skills in dialogue with AI.

Enterprise-grade The future of AI Assistant: From assistant to protagonist

Today's enterprise IT system is basically based on the architecture of "user interface + logic + data". Judging by the scenario of "talking to data", I think

The logic and flow parts will gradually " AI"

。 At present, AI plays more of an assistant role in enterprise IT systems, and in the near future, "assistant" will become the "protagonist".

No matter how "volatile" the current industry competition is, we should believe that science and technology can create a better life and future for mankind. IBM is also willing to communicate with more Chinese enterprises to jointly explore new technologies and models to enhance competitiveness.

IBM experts' view: Enterprises need to dispel the myth of AI applications, and "talking to data" will reshape the business model

IBM Corporation logo.

concerning IBM

IBM is the world's leading provider of hybrid cloud, artificial intelligence and enterprise services, helping customers in more than 175 countries and territories gain business insights, streamline business processes, reduce costs and gain a competitive advantage in their industries from the data they own. More than 4,000 government and enterprise entities in critical infrastructure sectors, including financial services, telecommunications, and healthcare, rely on IBM Hybrid Cloud Platform and Red Hat OpenShift for fast, efficient, and secure digital transformation. IBM's breakthrough innovations in artificial intelligence, quantum computing, industry cloud solutions and enterprise services provide our clients with open and flexible options. A long-standing commitment to corporate integrity, transparent governance, social responsibility, a culture of inclusion, and a spirit of service is the cornerstone of IBM's business. For more information, please visit:

Media Contact

CUI Shoufeng,