Urgent.News

What's breaking now, across thousands of outlets.

AI

The promise and peril of AI in government

In Taoyuan, a Taiwanese city with 200 semiconductor manufacturing facilities, Mayor San-Cheng Chang is utilizing AI to spearhead a transformative agenda. Holding expertise in both technology and government, Chang previously served as Taiwan’s first nonpartisan premier and in key leadership roles at Google and Acer. As the city’s director of Taiwan’s National Center for High-Performance Computing, Chang exemplifies the convergence of technology and governance that underpins Taoyuan’s AI integration.

Facing challenges such as heavy truck traffic, air pollution, rapid population growth, and constant construction, Taoyuan has harnessed AI to tackle these issues head-on. Since March 2023, the city's AI-powered lung cancer screening initiative, targeting residents over 40 with risk factors like occupational exposure and family history, has screened over 400,000 individuals, detecting 233 cases of lung cancer (90% in early stage) and yielding cost savings of NT$500 million ($17 million) through AI-aided diagnostics.

To manage the high volume of cases without overburdening local hospitals, Taoyuan has employed advanced AI software capable of predicting cancer risk from a single low-dose CT scan, eliminating the need for additional clinical data. This enables healthcare professionals to efficiently prioritize cases based on risk levels. Moreover, AI systems have been developed to address noise complaints about illegally modified vehicle exhaust pipes by analyzing exhaust pipe images against a database of legal configurations, resulting in a 20% reduction in such complaints compared to the previous year.

AI-powered cameras monitor construction sites and factory chimneys for pollution, automatically triggering water sprays when dust levels rise excessively or alerting inspectors to suspicious changes in emissions. Traffic-light cycles at busy intersections are optimized by AI, reducing wait times by approximately 10 seconds. Ambulances benefit from electronic fences that detect their presence, creating green-light corridors that cut average travel times from four minutes to just over two minutes, potentially saving lives.

Despite these successes, Mayor Chang remains cautious in his selection of AI projects. He refrains from implementing generative AI chatbots that may generate inaccurate information and carefully distinguishes between utilizing machine learning to detect existing conditions and using it to predict future outcomes. Taoyuan's AI applications focus on identifying patterns, such as spotting existing cancer or identifying vehicles with illegal modifications, without automating decision-making processes.

However, the trend among governments and companies is shifting towards using machine learning algorithms to predict future occurrences, such as recidivism risk, in making life-changing decisions ranging from sentencing and parole to credit scoring and job eligibility. Princeton University’s Arvind Narayanan cautions against the reliance on AI as a "Magic 8 ball," highlighting the grave risks associated with this approach.

He notes that job performance, criminal activity, or life expectancy predictions involve numerous variables and unpredictable events that historical data cannot fully capture, leading to potential discrimination and errors.

Written by urgent.news from Rest of World's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at elink32b.restofworld.org →

More in AI

South Korea-based DeepX tapes out DX-M2, an edge AI chip to run inference of 100B-parameter LLMs on robots and drones, targeting fall 2027 for mass production (Sally Ward-Foxton/EE Times)

DeepX is targeting edge markets including robotics with its second-generation AI accelerator, which can run large LLMs on device in under 5 W.

  • DeepX completes DX-M2 prototype with Samsung's PIM technology
  • DX-M2 enables inference of 20-100B parameter LLMs at 5W power
  • Targeted for mass production in fall 2027 for robotics and drones

More from Friday 13 November →