When I searched for "AI prospects" on Google, I got 57 million results. This shows how popular AI is becoming. The AI technology wave is not only "hot" in America but also spreading quickly to Vietnam. This opens up new chances for growth that we haven't seen before.
However, as legendary investor Warren Buffett famously warned: "Be fearful when others are greedy," the careless application of technology can lead to serious consequences. Here are two examples:
In 2013, the Dutch government launched an AI application called "SyRI" (System Risk Indication) with the aim of automatically detecting cases of tax and social welfare fraud. However, this system had fundamental flaws in its design, coupled with a lack of proper testing and error correction, which resulted in tens of thousands of families being wrongly labeled as fraudsters. SyRI pushed many households into bankruptcy and family breakdown, with some cases even leading to suicide. This scandal was exposed in the media in 2019, causing widespread outrage in Dutch society and leading to a large-scale investigation. Prime Minister Mark Rutte's government cabinet resigned en masse in January 2021 under public pressure. The SyRI system was shut down in 2020, but the process of compensating affected families continues to this day.
The second story involves AI errors in widespread medical diagnosis in the United States. Epic, a major electronic health records company, developed a tool to predict sepsis risk based on electronic clinical data. Epic's 2017 research report claimed their AI system had an accuracy rate of 83%, leading to its widespread implementation across hundreds of US hospitals. However, in 2021, an independent medical research team from the University of Michigan re-evaluated this system and discovered that the actual accuracy was only around 63%. When comparing the prediction results with the clinical conditions of 7,000 patients whom the AI identified as at risk of sepsis, only 12% actually had the condition. This serious consequence forced Epic to shut down their AI system in 2022.
I share these two examples not to deny the benefits that AI brings, but to emphasize the importance of being careful when implementing AI applications. Based on my years of experience in software design and working with AI systems, I believe that safe AI implementation needs to follow certain criteria.
I share these two examples not to deny the benefits that AI brings, but to emphasize the importance of being careful when implementing AI applications. Based on my years of experience in software design and working with AI systems, I believe that safe AI implementation needs to follow certain criteria.
First, we need independent and objective evaluations of AI system quality, comparing them with human performance under similar conditions. These evaluations should be done regularly to ensure system quality remains consistent when there are changes in AI algorithms or deployment environments.
Second, all AI software systems need clear warning mechanisms to alert users about potential errors and include features to collect user feedback. This feedback should be carefully analyzed to better understand the limitations of the AI system.
Third, AI software systems deployed in real-world situations should be designed to support humans, allowing them to maintain direct control and supervision. An interesting experiment showed that when a simple AI chess system worked together with an amateur chess player, they could defeat both the most advanced AI systems and top professional players. Designing AI systems to assist humans in their work, rather than completely replacing human roles, not only increases safety but can also improve final work outcomes.
For example, in a hospital setting, an AI system might help doctors by analyzing patient data and suggesting possible diagnoses, but the final decision would still rest with the medical professionals. Similarly, in manufacturing, AI can assist workers by identifying potential defects or suggesting process improvements, while workers maintain oversight of the production line. This human-AI collaboration approach ensures that we benefit from AI's computational power while maintaining human judgment and experience in critical decision-making processes.
When carefully researched and cautiously implemented, AI remains a promising technological trend that can bring positive changes to human life. Google's PageRank algorithm is a perfect example of how AI has completely transformed the way humanity accesses and shares knowledge on the internet. Reports from Google's Waymo self-driving cars also show that autonomous vehicles have nearly seven times fewer accidents compared to human-driven cars.
The year 2024 has witnessed a strong rise in technology investments in Artificial Intelligence (AI) in Silicon Valley. In September 2024 alone, venture capital funds invested a record-breaking $12.2 billion into AI-focused startups. This shows how much trust investors have in AI's future.
For instance, xAI, a company founded by Elon Musk less than two years ago, successfully raised $12 billion with a valuation reaching $50 billion. This valuation is equal to 500 times the company's annual profit, which is only about $100 million. This high valuation shows how much faith investors have in AI's potential. The huge gap between current profits and market value might seem surprising, but it reflects how strongly investors believe in AI's future growth. They think AI companies like xAI will bring big changes to how we live and work, even though right now these companies might not be making much money. This investment trend suggests that AI will continue to be a major focus in the technology sector for years to come.
Traditional tech giants like Google, Facebook (Meta), Microsoft, and Amazon are also actively participating in this race, with a combined investment of over $200 billion in AI research and development for 2024. Many analysts predict that this investment pace will continue to increase throughout 2025.
Facing this powerful AI wave, instead of ignoring or resisting this trend, we need to learn and improve our understanding of this technology to invest in it effectively and safely.
This article references content shared by Sayash Kapoor, an AI PhD researcher at Princeton University and author of "AI Snake Oil". Ned Nguyen.
--
Source: vnexpress - 02/01/2025 - https://vnexpress.net/ai-vang-thau-lan-lon-4834495.html
0 Comments