AI’s benefits, drawbacks, and safety concerns

@by Mark Ollig

A Sept. 17, 2025, Pew Research Center survey found that only 13% of Americans are comfortable receiving help from artificial intelligence, while 27% prefer no AI assistance at all.

Pew reported that 50% of Americans are more concerned than enthusiastic about AI’s growing role in daily life.

About 76% say it is important to know whether content is created by AI or a person, but 53% stated they lack confidence in being able to distinguish between the two.

On problem-solving skills, 29% believe AI assistance will help, and 38% think it will worsen our own analytical skills.

Many support AI in certain roles: 74% favor it in weather forecasting, 70% in detecting financial crimes, 70% in finding fraudulent checks, 66% in medical drug development, 61% in identifying crime suspects, and 46% in mental health support services.

According to another Pew report released April 3 of this year, 64% of US adults expect AI to reduce jobs over the next 20 years.

Only 5% expected more jobs, 14% expected little change, and 16% were unsure.

A Gallup survey found that the regular workplace use of AI increased from 21% in 2023 to 40% this year, while frequent usage rose from 11% to 19%.

Since 2011, Minnesota Information Technology Services (MNIT), has managed IT for the state executive branch, including infrastructure, software applications, and cybersecurity.

MNIT established statewide policies aligned with the One Minnesota Plan, appointed the state’s first AI director, and created an AI Leads group (a cross-agency team that coordinates AI work and helps agencies apply policy).

In 2023, it released the Public AI Services Security Standard.

According to the US National Institute of Standards and Technology (NIST), organizations that build or use AI should limit system access to authorized personnel and continuously monitor for bias or irregular performance.

The White House announced July 10 America’s AI Action Plan to boost innovation and adoption, simplify procurement, and strengthen protections for data centers, chips, cybersecurity, and responsible use.

A late-July directive instructed federal agencies to expedite environmental reviews and federal permit approvals for large data-center projects to shorten their completion timelines.

OpenAI’s ChatGPT now manages real-time phone conversations through its Realtime application programming interface (API).

Telephone service providers use Session Initiation Protocol (SIP) to set up the call, connect to the PSTN (public switched telephone network), and convert the caller’s audio into a data stream for the Realtime API.

AI processes the speech, generates a reply, and the provider sends it back to the caller as phone audio.

Because the SIP connection ties into the PSTN, a ChatGPT-powered agent can place and receive calls to regular phone numbers.

Meta (formerly Facebook) is building its largest data center in Richland Parish, LA, to support AI model training, with a cost of approximately $10 billion.

The data center is scheduled to come online later this decade.

Microsoft announced May 8, 2024, a $3.3 billion AI data center in Mount Pleasant, WI, expected to open in early 2026.

The data center will use thousands of NVIDIA graphics processing units (GPUs), which are specialized chips that perform multiple calculations simultaneously, enabling AI systems to be developed more quickly and respond faster when in use.

Microsoft’s Copilot AI is integrated into Word, Excel, PowerPoint, Outlook, and Visual Studio, with coverage expanding over time.

Minnesota IT Services organized the Transparent Artificial Intelligence Governance Alliance (TAIGA) in July 2023 to coordinate state AI policy, governance, and safety.

TAIGA helped publish the state’s Public Artificial Intelligence Services Security Standard in October 2023.

The Minnesota Department of Transportation has adopted a Generative Artificial Intelligence Standard (IT-003, effective July 14 of this year).

Generative AI refers to software that generates new content, such as text, images, audio, or code, based on patterns learned from existing data.

The city of Rochester operates the 311 phone number for non-emergency help; there, its AI-powered “Ask Chester” chatbot answers questions and takes service requests 24/7.

Anoka County is piloting an AI voice system for non-emergency calls, and Dakota County 911 is employing AI attendants.

The Mayo Clinic in Rochester uses AI to support diagnosis and research, including digital pathology and analysis of electrocardiogram (ECG) heart tests, to help clinicians make quicker, more precise decisions.

Telecommunications providers operating in Minnesota employ AI in their network operations to analyze traffic, enhance call routing, and improve reliability through traffic optimization, proactive maintenance, and self-healing networks that automatically correct certain faults.

T-Mobile is enhancing its 5G network with AI through a partnership with NVIDIA.

Verizon uses AI for energy efficiency and network optimization, and AT&T applies AI for automation and management of its network operations.

My advice: verify AI-generated content because AI can make mistakes. Ask it to cite its sources, and check them out.

Minnesotans and people across the country continue to debate the benefits, drawbacks, and safety concerns of AI’s growing role in our daily lives.