AT&T and GSMA are drawing attention to the real challenges of putting AI inside telecom networks. Mark Austin and Louis Powell explain that simply adding AI is not enough when systems handle critical infrastructure. Their discussions cover telco-trained models, smart routing, guardrails, and the cost of scaling AI agents across large networks. The message is that AI needs careful design, not just enthusiasm.
Meanwhile, the debate over how businesses should use AI is heating up. A MIT study found that 95 percent of organizations see no return on their AI investments, yet AI-linked stocks now make up 45 percent of the S&P 500 market cap. A former defense AI expert argues that simpler tools often outperform complex models. He urges companies to match the technology to the task rather than chase the latest platforms, warning that pension savers and workers could suffer if an AI bubble bursts.
On the healthcare front, Luminarity founder Shruti Kothari is set to speak at an event about balancing AI innovation with cybersecurity in medicine. The conversation will focus on responsible AI governance and how to keep systems secure while still pushing forward. Separately, a new editorial in Open Heart shows how AI is already making its way into cardiovascular care. Tools like ChatGPT-4o and XGBoost are being tested to predict heart risks and interpret angiograms, while explainable AI methods help doctors understand how risk scores are calculated.
Network infrastructure is another major concern. AI workloads demand high bandwidth and low latency to move data between GPUs effectively. Adding more compute power will not solve problems if the underlying network is too slow. CIOs are being urged to consider network limits alongside processing power to avoid costly bottlenecks. Separately, Jim Himes has suggested that AI data centers could eventually be designated as critical infrastructure in the United States, raising new questions about resilience and security.
AI is also showing up in unexpected places, from music to classrooms. Spotify has seen a rise in AI-generated music projects, with artists like George Lewis and Dadabots exploring what machine-made sound can be. In schools, one teacher used AI to create a photo of a six-year-old student as a grown-up, sparking concern from her mother. The Charleston County School District said it is preparing AI literacy training for teachers. Martha Minow also spoke at MIT about how AI can both challenge and support democracy, calling for responsible use.
Key Takeaways
- AT&T and GSMA stress that AI in telecom requires guardrails, telco-trained models, and smart routing, not just deployment
- 95 percent of organizations see no return on AI investments, while AI-linked stocks make up 45 percent of the S&P 500 market cap
- A former defense AI expert recommends practical tools over complex models to avoid an AI bubble
- Luminarity founder Shruti Kothari will discuss responsible AI governance and cybersecurity in healthcare
- ChatGPT-4o and XGBoost are being tested in cardiovascular medicine for risk prediction and angiogram interpretation
- AI workloads need strong networks with high bandwidth and low latency; adding GPUs alone will not fix bottlenecks
- Jim Himes says AI data centers could eventually be named critical infrastructure in the US
- Ghost AI, or abandoned AI tools with lingering access, poses a security risk that requires ongoing governance
- Charleston County School District is preparing AI literacy training after a teacher used AI to generate a student photo
- Martha Minow spoke at MIT about sustaining democracy through responsible AI use and overcoming its risks
AT&T and GSMA tackle AI challenges in telecom networks
AT&T's Mark Austin and GSMA's Louis Powell discuss how to deploy AI in telecom networks. They stress that AI needs special care inside critical infrastructure. Key topics include telco-trained models, smart routing, guardrails, and the economics of scaling AI agents. The conversation highlights that simply adding AI is not enough.
AT&T and GSMA share what it takes to run AI in telecom
AT&T and GSMA explain the real work behind putting AI into telecom systems. Mark Austin and Louis Powell cover telco-trained models and smart routing. They also talk about the need for guardrails and the cost of scaling agents across the network. The episode offers a deeper look at building AI for critical infrastructure.
Healthcare event explores AI innovation and cybersecurity balance
Luminarity founder Shruti Kothari will speak at an event about responsible AI governance in healthcare. The discussion will cover how to balance innovation with secure and reliable systems. Attendees will learn about AI challenges in healthcare and best practices. The event brings together healthcare professionals, AI experts, and cybersecurity specialists.
Exploring the creative side of machine-made music
Spotify has seen a rise in AI-generated music projects. One project acknowledged its AI origins after gaining attention. Trombonist George Lewis has collaborated with AI since the 1980s. Dadabots uses SampleRNN technology to create music and faces challenges in finding the best parts. The article also recommends human-made music for listeners seeking a different experience.
AI performance depends heavily on the underlying network
AI workloads need strong networks to move data between GPUs effectively. Adding more GPUs will not fix problems if the network is too slow. AI traffic demands high bandwidth and low latency, especially during training. CIOs must consider network limits alongside compute power. Understanding infrastructure limits helps prevent costly bottlenecks in AI operations.
Businesses Should Choose Practical AI Over Complex Models
Companies are spending billions on advanced AI even though a MIT study found 95 percent of organizations see no return. AI-linked stocks now make up 45 percent of the S&P 500 market cap. A former defense AI expert argues that simpler tools often work better than complex ones. He says businesses should match AI to the task instead of chasing the latest technology. Pension savers and workers could be hurt if the AI bubble bursts.
Mom Objects After School Uses AI to Create Daughter's Photo
Nikki Anderson opened her 6-year-old daughter Harper's Friday folder and found an AI-generated photo showing Harper as a 25-year-old nurse. The teacher used AI to create the image, which Harper liked because of the wavy hair. Anderson is upset because she believes her child's own imagination was replaced by an algorithm. The Charleston County School District said it is preparing AI literacy training for teachers. A professor said teachers should not put student personal information into AI models.
AI Improves Heart Care Through Better Diagnostics and Risk Prediction
A new editorial in Open Heart explains how AI is advancing cardiovascular medicine. Deep learning can turn chest X-rays into tools for detecting mitral stenosis. Machine learning helps spot high-risk plaque in coronary CT scans. Tools like XGBoost and ChatGPT-4o are being tested to predict heart risks and interpret angiograms. Explainable AI methods such as Shapley explanations help doctors understand why a risk score was given. The authors say AI must be integrated into daily clinical work with doctor oversight.
Ghost AI Poses Security Risk After Tools Are Approved
Ghost AI refers to approved AI tools, agents and automations that keep accessing business systems long after they are needed. Employees may leave or projects may end, but the tools still have permissions and no clear owner. Security teams should track which AI tools can access company data and whether that access still matches the original purpose. The article says AI governance is not a one-time approval but an ongoing responsibility. Applying the principle of least privilege helps reduce unnecessary exposure.
AI Data Centers May Be Named Critical Infrastructure in US
Jim Himes, ranking member of the House Permanent Select Committee on Intelligence, said AI data centers could eventually be designated as critical infrastructure. He spoke at an October 21 event at the Brookings Institution in Washington, D.C. He said the US needs a more robust and distributed computing infrastructure with redundancy. The United States currently recognizes 16 critical-infrastructure sectors, and AI data centers are not yet a standalone sector. The shift raises questions about data security, resilience and what happens when AI infrastructure fails.
Martha Minow discusses AI and democracy at MIT lecture
Martha Minow, the Henry N. Smyth Professor of Law at Harvard Law School, delivered the Compton Lecture at MIT on Oct. 23. Her talk focused on sustaining democracy through the use and overcoming of AI. The lecture addressed how artificial intelligence can both challenge and support democratic systems. Minow explored ways society can use AI responsibly while addressing its risks.
Sources
- The Five Nine: How AT&T and GSMA are building AI for telecom networks
- The Five Nine: How AT&T and GSMA are building AI for telecom networks
- Event: Balancing healthcare AI innovation and cybersecurity
- Beyond AI slop: The strange art of machine-made music
- Your AI is only as good as the network underneath it
- Companies Need Right-Sized AI, Not Maximum Complexity
- Her 6-year-old brought home an AI-generated photo from school. This mom isn’t having it
- Artificial Intelligence Advances Precision Cardiovascular Care
- Ghost AI: The Security Risk Businesses May Be Forgetting
- AI Data Centers Could Become Critical Infrastructure as Federal Security Concerns Grow
- At MIT, Martha Minow speaks on sustaining democracy, both by using and overcoming AI
Comments
Please log in to post a comment.