Technology

MIT study finds AI has the effect of limiting tasks that require visual stimulation.

2024-01-29 03:07:23

Researchers from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) worked with IBM's Institute for Business Value to assess the potential impact of machine learning algorithms and AI services on the job market. They found that full algorithmic systems will not be economically viable for most in the next decades.

This study focuses on the desired work. "Visual analytics," such as product inspection for quality control at the end of the production line, researchers determined that only 23 percent of wages paid for vision work are economically attractive to automation. Humans will continue to be the better economic choice for these jobs.

Goldman Sachs has previously predicted that 25 percent of the entire job market will be replaced by AI automation in the next few years, while McKinsey has stated that half of the world's workforce will be powered by AI by 2055. Researchers at MIT suggest that AI should experience Gradual integration across sectors Even in fields such as computer vision. which the cost of training the model has been seen as "Important progress"

MIT CSAIL scientist Neil Thompson said his team found "Significant potential" for automation-driven work AI However. When considering their overall cost, many of these tasks are not yet "interesting" for automation. AI despite costs dropping 20 percent every year.

The researchers acknowledge some limitations to their study. Because chatbots are not covered. Considering the self-hosted services sold by OpenAI and other vendors, there are text and image generation models and other next-generation AI services currently on the market, such as ChatGPT and Midjourney. That only requires fine-tuning to automate specific tasks. Researchers suggest that many low-wage human jobs will still be more economically viable than full AI automation.