As artificial intelligence (AI) increasingly enters the industrial sector, many companies face challenges in assessing the authenticity and effectiveness of AI technologies from suppliers. On March 14, at the Yuexiu iPARK Guangdong-Hong Kong Smart Valley, Guangzhou, the industrial AI technology company Mogulinker held an event focused on "debunking" industrial AI. They highlighted common types of misleading AI applications, such as rule-based fake AI, data-masking AI, black-box AI, and function-bridging AI. Companies must carefully discern the authenticity of industrial AI to avoid hindering technological innovation and ecosystem development.
Traditional Software 'Masquerading' as AI
Mr. Huang, a representative from a pharmaceutical company, shared with the Wen Wei Po that his factory's engineering department has been exploring digital systems for some time. As a leader in digital transformation, they frequently encounter numerous AI suppliers and find it challenging to verify the authenticity of AI solutions.
"Some suppliers make grand claims about their industrial AI products, yet they often lack targeted AI solutions for common industrial applications. This can easily mislead those without relevant expertise," he said.
The prevalence of "fake AI" can lead to financial losses for companies and significantly delay innovation and competitiveness.
Expert Insight on AI Validation
Shen Guohui, founder and CEO of Mogulinker, surveyed 100 industrial practitioners. He found that 77% of respondents identified the verification of AI effectiveness as a significant barrier to implementing industrial AI applications. Many companies expressed the need for professionals with keen insight to help distinguish real AI from fake.
Shen stated, "Any AI that cannot be visualized is unreliable." He noted that the most apparent feature of "fake AI" is its reliance on jargon-laden PowerPoint presentations, where actual AI benefits are not experienced by customers. He admitted that his company once included unlaunched AI technology in a presentation to expedite software sales, which turned out to be nothing more than a collection of concepts and buzzwords—an example of "fake AI." Many suppliers continue to promote their AI technologies, which are essentially traditional software disguised as AI.
Developing Solutions to Identify Genuine AI
To combat the circulation of "fake AI," Mogulinker has developed a vertical large model for industrial auxiliary energy systems. This model can understand over 20 types of high-energy-consuming equipment and features capabilities in inquiry, diagnosis, prediction, and control, effectively helping manufacturing enterprises quickly verify the authenticity of AI applications. This model has been recognized as a typical application case by the Ministry of Industry and Information Technology of China.
Dr. Zhou Ziye, an overseas high-level talent under Shenzhen's "Peacock Plan" and a doctoral candidate in Systems Engineering and Engineering Management at the Chinese University of Hong Kong, stated in an interview that industrial enterprises should not pay for AI hype but should invest in value. True AI should be implementable, capable of creating measurable and sustainable value, and empower high-quality industrial development.
(Source: Wen Wei Po)
Related News:
Dobot Robotics launches embodied intelligence robots: Stock price surges over 40%
Comment