AI is moving into more real-world settings, but the articles emphasize that deployment depends on infrastructure, oversight and practical limitsânot just powerful models. At North Dakota State University, Xin âRexâ Sun is developing a sensor-equipped system that can diagnose crop problems and support decisions directly on farm machinery. His team uses Nvidia GPUs for demanding research and smaller systems such as Raspberry Pi and Arduino elsewhere. Because tractors and combines may lack reliable internet and must also navigate and avoid obstacles, computing has to fit their power, heat, cost and safety constraints.
Enterprise systems face a related infrastructure challenge. AI agents may run code, query databases and use business applications, so production systems need CPUs, storage, networking and security alongside GPUs. Vultr says its integrated cloud platform includes GPU computing and access to more than 60 Cloud Native Computing Foundation projects. VAST Dataâs Brian Verkley describes a CAMP stackâcoding agents, an agent runtime, model services and a platformâthat can let teams test tools while keeping data in place. His example combines S3 storage, NFS access, event-driven processing and SQL tables for vectors and metadata; teams still need to manage cost, database issues and security.
Governance and human judgment also emerge as recurring concerns. An article on military biodata urges the Pentagon and Congress to establish rules for collecting and using signals such as heart rate, sleep, temperature and blood chemistry. Such data may support training and care, but readings are not diagnoses, and AI readiness scores should not automatically dictate command decisions. The article warns that combined biodata could expose privacy risks or unit vulnerabilities. In HR, Macquarie Business School researcher Dr. Miles Yang says automated hiring, workforce planning and agentic tasks such as shift allocation can miss illness, caregiving and other changing circumstances. He calls excessive reliance on rigid systems an âefficiency trapâ; the article cites job cuts at National Australia Bank, Atlassian and Salesforce, and AI customer-service agents from Australian startup Lorikeet.
Communities are weighing the costs of the infrastructure behind AI. At a University of Dubuque forum, students and journalists discussed data centersâ possible effects on water, energy, land and the environment, as well as ethical and policy questions. English professor Donavan Tann said communities should help shape technologyâs role in their future. NAACP president Derrick Johnson separately warned that data center expansion can displace residents and damage homes and businesses in historically marginalized U.S. communities, sometimes without fair compensation. The NAACP is calling for transparency, accountability and inclusive decision-making.
Other projects focus on specific applications and skills. UCLA researchers are refining an early-stage nanotechnology and AI method that analyzes bodily-fluid samples for drugs including cocaine, heroin and methamphetamine in a few minutes. They say it may cost less and produce fewer false positives than traditional screening, and hope for commercial use within the next few years. The College of American Pathologists and the University of Pittsburghâs CPACE are developing pathology AI tools, including context-aware Nebulon-GPT-Plus (Pitt-GPT-Plus) and the no-code Pitt-AI-cademy, with human oversight, security and responsible governance as priorities. The University of Phoenix has also built an MBA framework that integrates AI skills with core business learning, using employer input and labor-market data; ACBSP Region 7 Chair Jim Lipot praised its workforce focus.
For companies adopting AI, Cognitionâs chief revenue officer advises starting with a clearly understood customer problem and a credible value proposition. Data quality and availability, trust and customer relationships matter, and adoption should be strategic rather than merely tactical. The broader message across these developments is that useful AI depends on matching technology to the task while keeping people, security and affected communities involved.
Key Takeaways
- North Dakota State Universityâs Xin âRexâ Sun is developing on-machine crop diagnostics, using Nvidia GPUs as well as Raspberry Pi and Arduino systems suited to different farm-machine constraints.
- Production AI agents need CPUs, storage, networking and security in addition to GPUs; Vultr says its platform includes access to more than 60 Cloud Native Computing Foundation projects.
- VAST Dataâs proposed CAMP stack lets coding agents work across data and tools without requiring teams to move their data, but cost, security and database management remain necessary.
- An article urges the Pentagon and Congress to govern military biodata use, warning that sensor readings are not diagnoses and AI readiness scores should not automatically drive command decisions.
- Macquarie researcher Miles Yang warns that AI-driven HR can miss individual circumstances; the article cites job cuts at National Australia Bank, Atlassian and Salesforce and AI customer-service agents from Lorikeet.
- At a University of Dubuque forum, participants examined data-center effects on water, energy, land and the environment; NAACP president Derrick Johnson calls for transparency and inclusive development amid displacement concerns.
- UCLA researchers are refining an AI-and-nanotechnology drug-screening method that tests bodily fluids for cocaine, heroin and methamphetamine in minutes, with commercial use hoped for within a few years.
- The College of American Pathologists and University of Pittsburghâs CPACE plan pathology AI work using Pitt-GPT-Plus and the no-code Pitt-AI-cademy, emphasizing human oversight and data security.
- The University of Phoenixâs ACBSP-recognized MBA framework adds practical AI skills while retaining core business learning; Cognitionâs sales chief stresses clear customer value, data quality and trust.
Edge AI brings crop decisions directly to farm machinery
Xin âRexâ Sun of North Dakota State University is developing a sensor-equipped platform to help diagnose crop problems and support actions in the field. He says agricultural AI needs to run on robots, combines, and tractors, especially where internet access is unreliable or machines must respond quickly. His team uses Nvidia GPUs for demanding research and smaller systems such as Raspberry Pi and Arduino for other tasks. Farm machines increasingly need to combine crop analysis with navigation, obstacle detection, and control, so developers must match computing power to the machineâs power, heat, cost, and safety limits.
Enterprise AI needs more than GPU power
As companies move AI agents into production, they need more than GPUs to train and run models. Agents may also execute code, access databases, use business applications, and connect to external tools, requiring CPUs, storage, networking, security, and safe workspaces. Vultr says it offers an integrated platform with cloud and GPU computing, networking, storage, and access to more than 60 Cloud Native Computing Foundation projects. Cochrane said CPUs remain important for databases, application code, data pipelines, and some smaller models. The article also highlights data sovereignty and secure execution as growing needs for enterprise AI infrastructure.
Military leaders must govern how AI uses biodata
As AI and biotechnology come together, the article argues that the Pentagon and Congress must set rules for collecting, sharing, interpreting, and using military biodata. Sensors can measure heart rate, sleep, temperature, blood chemistry, stress, and other signs that may help with training, recovery, medical care, and readiness planning. But a sensor reading is not a diagnosis, and an AI readiness score should not become an order by default. The author says human judgment and accountability must remain central to command decisions. The article also warns that biodata can expose unit vulnerabilities or create privacy and security risks, especially when it is combined with other military information.
Dubuque forum weighs AI and data center impacts
The University of Dubuque held a public forum on Tuesday night about the possible benefits and drawbacks of AI and data centers. Students examined effects on water use, energy demand, land, and the environment, along with ethical and public policy questions. The event included student presentations and a 90-minute discussion and question session with journalists Benjamin Fisher, Tom Kertscher, and Grace Nieland. English professor Donavan Tann said communities should help decide how the technology shapes their future. Student Tate Glaspie said people should weigh possible improvements in quality of life against potential costs.
AI could make HR faster but less flexible
Macquarie Business School researcher Dr. Miles Yang warns that relying too heavily on AI for human resources decisions can make companies less able to respond to individual circumstances. Businesses already use AI to support hiring and workforce planning, while agentic AI could also allocate shifts, monitor performance, or trigger employment processes. Yang says algorithms may apply rules consistently but can miss factors such as illness, caregiving, or unexpected crises. He calls this the âefficiency trap,â in which rigid automated processes lead to poor decisions when situations change. The article cites job cuts at National Australia Bank, Atlassian, and Salesforce, as well as AI customer service agents developed by Australian startup Lorikeet.
University of Phoenix MBA AI framework earns ACBSP recognition
The University of Phoenix developed a data-driven framework for adding artificial intelligence skills to its Master of Business Administration curriculum. A white paper describes how the framework connects AI skills to existing learning goals while protecting core business knowledge and critical thinking. Students use AI tools in realistic business situations and create portfolio-ready work that shows practical skills. The approach uses faculty expertise, employer input, and labor market data, and has expanded to other business programs. ACBSP Region 7 Chair Jim Lipot praised the framework for helping educators prepare students for workplaces shaped by AI.
UCLA researchers develop AI and nanotechnology drug screening
Researchers at the University of California, Los Angeles developed an early-stage technology that uses nanotechnology and artificial intelligence to screen for illicit drugs on site. Nanoparticles are applied to bodily fluid samples, such as blood or urine, and AI analyzes the results for drugs including cocaine, heroin, and methamphetamine. The article says the process takes a few minutes and may cost less and produce fewer false positives than traditional screening. Researchers are refining the technology and working to scale up production, while exploring possible use in law enforcement and addiction treatment. The article says the team hopes it will be ready for commercial use within the next few years.
Cognition sales chief shares lessons on selling AI to enterprises
Cognition's chief revenue officer shared advice on enterprise AI sales in an interview with Grit, the venture arm of Kleiner Perkins. He said sales teams should understand the customer's problem and explain how AI can address it with a clear value proposition. He also highlighted data quality and availability as key to AI adoption. Building customer relationships and trust matters, and companies should take a strategic approach to adopting AI rather than a tactical one.
NAACP warns AI data centers may harm marginalized US communities
NAACP president Derrick Johnson warned that the growth of AI data centers is affecting historically marginalized communities in the United States. He said data center expansion can displace residents and destroy homes and businesses, while communities may not receive fair compensation for their land. The NAACP is calling for greater transparency, accountability, and inclusive decision-making in data center development. In an interview with Anadolu, Johnson also stressed the need to address systemic inequalities that have concentrated data centers in these communities.
CAP and University of Pittsburgh team up on AI tools for pathology
The College of American Pathologists and the University of Pittsburghâs Computational Pathology and AI Center of Excellence will work together to develop AI solutions for pathology. Their plans include improving efficiency, knowledge management, education, and access to CAP resources. Two CPACE platforms will support the work: Nebulon-GPT-Plus, also known as Pitt-GPT-Plus, for context-aware AI assistance, and Pitt-AI-cademy, a no-code platform for AI education. Possible uses include finding CAP educational resources, supporting administrative work, and accessing information on laboratory quality and accreditation. The organizations say they will focus on careful assessment, human oversight, data security, and responsible governance.
Coding Agents Could End the Cloudâs Old Tradeoff
Brian Verkley, director for AI data platforms at VAST Data, says coding agents could help developers try new tools without moving their data or committing to a new system. His proposed CAMP stack includes coding agents, an agent runtime, model services and a platform. In his example, photographs, videos and manuals stay in S3 while older apps access the same objects through NFS; events trigger processing, and SQL tables store vectors and metadata for searches. The agent runtime can find instructions and call a work order API, while teams can replace a vision model without moving the data. Verkley also notes that teams still need to manage database problems, costs and security, and that coding agents do not take over on-call duties.
Sources
- Edge AI Moves Agricultural Intelligence from the Cloud to the Field
- Enterprise AI infrastructure moves beyond GPUs
- Biodata, AI and the future of command: Getting it right
- University of Dubuque forum explores benefits, risks of AI and data centers
- What happens when AI takes the human out of human resources?
- University of Phoenix AI integration framework for MBA curriculum earns ACBSP recognition
- Nanotechnology and AI enable rapid on-site illicit drug screening
- Dealroom.co | Past $1B run rate with 80 reps: Cognition's CRO on bringing enterprise sales to AI
- INTERVIEW - AI boom reviving âsacrifice zonesâ in historically marginalized US communities: NAACP
- CAP and University of Pittsburgh Partner to Develop AI Tools for Pathology
- Coding Agents Break Cloud's Old Tradeoff
Comments
Please log in to post a comment.