Why did the American artificial intelligence Meta project "die in the womb"?

In today's era of rapid technological development, artificial intelligence has become a global focus, and many technology giants have invested a lot of resources in the hope of taking the lead in this field. Meta (formerly Facebook) has a long history of layout in the field of artificial intelligence. Over the years, it has continuously invested resources to attract top talents, build a large artificial intelligence research team and laboratory, and is committed to promoting the advancement of artificial intelligence technology. However, Meta's series of projects in the field of artificial intelligence have recently suffered major setbacks, and many high-profile AI products have "failed" and fallen into the dilemma of failure.

Take the AI ​​character project launched by Meta in September 2023 as an example. The core of the project is to create a chatbot using celebrity portraits, such as imitating the image of football star Tom Brady to create an "outspoken sports debater" Bru. In order to obtain the celebrity portrait rights, Meta invested up to $5 million. But contrary to expectations, the interactive effects of these AI characters are extremely awkward and weird, far from the expected interesting and vivid experience. In the end, this project, which cost a lot of manpower, material resources and financial resources, ended after only one year. Although Meta did not clearly explain the specific reasons for the failure of the project, it can be inferred from its statement that it learned how to make AI better serve users. The AI ​​image generation tool "Imagine" launched by Meta also has many problems. In order to compete with Google's AI model Gemini, Meta launched "Imagine with Meta AI" at the end of last year. Users enter text prompts and the tool can convert text into four realistic images. However, Mia Sato, a reporter for the technology news website The Verge, tried to use Meta Imagine to create images of Asian men and women paired with white people after "dozens" of attempts, but all ended in failure. Business Insider also tried to instruct Meta Imagine to generate an image of a Chinese husband and a white wife, but what appeared in the end was a picture of an Asian couple. The exposure of this major flaw has caused Meta Imagine to be questioned for racial bias. After all, Priscilla Chan, the wife of Meta CEO Mark Zuckerberg, is Chinese, and this embarrassing situation has further intensified speculation that Meta AI technology has racial bias. In fact, AI image generation tools such as Gemini and Imagine generate images by collecting a large amount of data and then calculating. If there are few photos of couples of different ethnicities stored, it is difficult to create corresponding images, which makes it easy to fall into racial bias. AI ethics expert Nakeem A. Stefflbauer once pointed out: "With the popularization of predictive algorithms and AI technology, it is difficult for the public to detect that these images are just immediate responses to general opinions, stereotypes or lies." In addition, Meta's layout in the field of AI is not limited to these two projects, and it has also made a lot of investment and exploration in many other aspects. In the field of social media, Meta attempts to use AI technology for content recommendation, user portraits and social relationship analysis, hoping to provide users with more personalized content recommendations by analyzing user behavior and interests, and improve user participation and satisfaction. In the field of virtual reality and augmented reality, Meta also hopes to use AI technology to bring users a more realistic experience. For example, in the virtual reality devices it develops, AI technology is used to identify users' movements and expressions to achieve more natural interactions. In the medical field, Meta also hopes that AI technology can assist doctors in disease diagnosis, drug development, and other tasks, and improve medical efficiency and quality. However, from the current actual situation, these beautiful visions have not been realized.

So, what exactly caused Meta's AI projects to fail one after another? From a technical perspective, although Meta has in-depth research and application in deep learning algorithms, and has achieved some results in fields such as image recognition and natural language processing, there are still many technical bottlenecks in actual product applications. For example, in the AI ​​role project, although huge investment has been made, it may not have made breakthroughs in key technical links such as emotional interaction and semantic understanding, resulting in unsatisfactory interaction effects of chatbots. In the Imagine tool, there may be defects in data collection and processing, which cannot cover sufficiently rich and diverse data, resulting in problems in generating specific ethnic matching images. From the perspective of market feedback, users' expectations for AI products are increasing day by day. They not only require powerful functions, but also require natural, real and unbiased experiences. Meta's AI products failed to meet users' needs in this regard, resulting in low user acceptance and satisfaction with its products. Taking the AI ​​character project as an example, the awkward interaction effect makes it difficult for users to generate interest in continuous use; and the racial bias questioning of the Imagine tool has seriously affected its image in the minds of users.

Meta's failure in the field of artificial intelligence has undoubtedly sounded the alarm for the entire technology industry. This shows that even technology giants with huge resources and top talents are not smooth sailing on the road of exploring artificial intelligence. In the future, if Meta wants to rise again in the field of AI, it needs to deeply reflect on the reasons for its failure, increase investment in technology research and development, break through technical bottlenecks, and pay more attention to user needs and market feedback to ensure that AI products can truly bring value to users. For other technology companies, Meta's lessons are also worth learning from. When investing in the field of AI, they need to plan carefully and fully consider multiple factors such as technology, market, and ethics to avoid repeating the same mistakes.