Research across several fields points to a practical theme: AI can be useful, but results depend on the task, safeguards and human oversight. A review of 160 quantum machine-learning studies published from 2015 to June 2025 found no clear, repeatable quantum advantage on practical tasks. In their own tests across five public datasets, hybrid quantum classifiers generally matched or slightly outperformed classical models while using fewer parameters.
A separate review of 80 studies on AI in enterprise resource planning, published from 2014 to 2024, identified machine learning, decision support, process optimization, predictive analytics and business intelligence as major themes. It also highlighted gaps in real-world testing, tools for small and medium-sized businesses, explainability and integration with older systems. In manufacturing, a Rootstock Software survey found 77% of respondents plan to increase AI spending or resources in the next 12 to 18 months; more than half expect increases of 10% to 25%. Of manufacturers using AI, 31% already use agentic AI, while 88% see value in one AI agent that works across business applications.
Security and responsible use remain central concerns. Microsoft Chief Digital Safety Officer Mike Jackson said at the 2026 Paris Peace Forum that AI can intensify risks such as bullying and exploitation, and called for safety to be built into products from the start. He favored helping children participate safely through technology, education and empowerment rather than banning them from social media, and discussed Microsoft’s Behind the Chat initiative. Separately, guidance on enterprise security urges organizations to map AI-agent identities and access alongside cryptographic dependencies. NIST finalized three post-quantum cryptography standards in August 2024—FIPS 203, 204 and 205—and organizations may need years to replace vulnerable cryptography.
Public-sector guidance likewise stresses human responsibility. An implementation package discussed by Italy’s Council of Ministers on 10 June 2026 promotes innovation while protecting fundamental rights. Training for civil servants should cover checking AI outputs, identifying errors, explaining decisions and knowing when to seek human review. Novi Community School District, which adopted an AI plan in August, aims to explore classroom uses while ensuring the technology does not do students’ thinking for them. A mid-career AI guide targets professionals already using AI who are ready to take on higher-stakes work.
One forecasting study offers a caution against assuming more complex models perform better: for Indonesian rice prices, simply carrying the current month’s price forward beat eight other methods over 48 forecast periods from January 2021 through December 2024. Its bootstrap mean absolute percentage error was 1.847%. The researchers used 84 monthly national records from January 2018 to December 2024 and found that forecast accuracy did not necessarily correspond to the quality or stability of explanations. The Financial Times, meanwhile, named Cristina Criddle its chief AI correspondent; she is based in San Francisco and joined the newspaper in late 2021.
Key Takeaways
- A review of 160 quantum machine-learning studies found no clear, repeatable practical quantum advantage; tested hybrid models generally matched or slightly outperformed classical models with fewer parameters.
- An ERP review of 80 studies identified five leading themes and gaps in real-world validation, SME-focused systems, explainability and legacy-system integration.
- Rootstock Software reports 77% of manufacturers plan to increase AI investment or resources in the next 12 to 18 months; 31% of AI users already use agentic AI.
- Microsoft safety chief Mike Jackson called for child safety to be built into AI products and discussed Microsoft’s Behind the Chat initiative.
- NIST finalized post-quantum standards FIPS 203, FIPS 204 and FIPS 205 in August 2024; organizations are urged to inventory identities, access and cryptographic dependencies.
- Guidance for public workers emphasizes AI training in output checking, error detection, decision explanations and human review; Novi schools also plan classroom AI use with safeguards.
- For Indonesian rice prices, a persistence forecast produced a 1.847% bootstrap mean absolute percentage error and beat eight other methods over 48 forecast periods.
- The Financial Times named Cristina Criddle its chief AI correspondent, based in San Francisco.
- A mid-career AI guide is aimed at professionals who already use AI and want to take on higher-stakes work.
Hybrid Quantum Models Match Classical Systems Without Proven Advantage
Researchers Raed Abu Zitar, Murad Al-Rajab, Samia Loucif, and Ahmad Nasayrah reviewed quantum machine learning studies published from 2015 to June 2025. They screened 5,463 publications and analyzed 160 studies using real or simulated quantum processors. The studies covered areas such as image classification, cybersecurity, finance, chemistry, and healthcare. In their own tests, the researchers compared five quantum classifiers with two classical classifiers across five public datasets, using five random seeds. Hybrid systems generally matched or slightly outperformed classical models while using fewer parameters, but the review found no clear, repeatable quantum advantage across practical tasks.
Review Maps How AI Is Changing Enterprise Resource Planning
A systematic review in the Journal of Big Data examined how artificial intelligence is being used in enterprise resource planning systems. The researchers analyzed 80 Scopus-indexed studies published from 2014 to 2024, using PRISMA-guided coding, bibliometric analysis, and science mapping. They identified five main themes, including machine learning, decision support, process optimization, predictive analytics, and business intelligence. The review found gaps in real-world testing, systems designed for small and medium-sized businesses, explainability, and integration with older systems. The authors proposed research to help move intelligent enterprise systems from prototypes toward explainable, connected tools.
Cristina Criddle Named Financial Times Chief AI Correspondent
The Financial Times has named Cristina Criddle its chief artificial intelligence correspondent. Based in San Francisco, Criddle has worked at the Financial Times since late 2021, covering technology companies and industry developments. Before that, she reported on technology for BBC News and also worked as a producer and reporter. Her earlier roles included jobs at The Telegraph, Sky, the Evening Standard, HuffPost, and ITV. She helped establish The Tab Liverpool in 2012 and graduated from the University of Liverpool.
Microsoft Safety Chief Calls for Safer AI Products for Children
At the 2026 Paris Peace Forum, Microsoft Chief Digital Safety Officer Mike Jackson spoke with GZERO’s Tony Maciulis about protecting children online as AI develops. Jackson said AI can intensify risks such as bullying and exploitation, and argued that companies should build safety into products from the start. He said efforts should focus on helping children participate safely online through better technology, education, and empowerment rather than banning them from social media. Jackson also discussed Microsoft’s Behind the Chat initiative, regulation, and the value of including young people’s experiences in online safety work. GZERO Media presented the conversation in partnership with Microsoft.
Simple Persistence Forecast Beats Machine Learning on Indonesian Rice Prices
A study in the Journal of Agriculture and Food Research compared nine forecasting methods for Indonesian rice prices. Across 48 forecast periods from January 2021 through December 2024, a simple method that carried the current month’s price forward had the lowest error, with a bootstrap mean absolute percentage error of 1.847 percent. The researchers used 84 monthly national records from Badan Pusat Statistik covering January 2018 to December 2024 and designed their process to prevent future data from leaking into forecasts. They also found that a model’s forecasting accuracy did not necessarily match the quality or stability of its explanations. Accurate rice forecasts help agencies such as the National Food Agency and Badan Urusan Logistik plan actions including buffer-stock releases and consumer subsidies.
Novi schools plan classroom AI use with strong safeguards
In August, the Novi Community School District adopted a plan to make Novi a state and national leader in artificial intelligence. Superintendent Ben Mainka discussed the plan with The Metro’s Sam Corey. The district wants to explore how AI can support students while setting limits so the technology does not do their thinking for them.
Manufacturers increase AI investment as agentic tools gain ground
A Rootstock Software survey found that 77% of manufacturers plan to increase AI spending or resources in the next 12 to 18 months, and more than half expect increases of 10% to 25%. Among manufacturers using AI, 31% already use agentic AI, which can plan and complete tasks with little human help. Overall, 64% have moved beyond exploring AI, while 61% are at least somewhat concerned that AI could limit career opportunities. Also, 88% see value in a single AI agent that works across business applications, and ERP remains among the top four technologies expected to shape AI adoption.
AI and quantum threats put enterprise security under pressure
AI agents are creating more identities, permissions and credentials for security teams to track, while future quantum computers could threaten public-key systems such as RSA and elliptic-curve cryptography. The article says organizations should map agent identities, delegated access and cryptographic dependencies so they can manage both risks. NIST has encouraged organizations to prepare for a shift to post-quantum cryptography and finalized three standards in August 2024: FIPS 203 for ML-KEM, FIPS 204 for ML-DSA and FIPS 205 for SLH-DSA. Planning is important because finding and replacing vulnerable cryptography across products and systems can take years, and stolen encrypted data could be stored for later decryption.
Public workers need training to use AI responsibly
Public managers and civil servants need training that helps them use AI while keeping human judgment and responsibility central, the article says. Italy’s Council of Ministers discussed an implementation package on AI use on 10 June 2026, based on promoting innovation while protecting human responsibility and fundamental rights. Training should go beyond using tools to include checking AI outputs, spotting errors, explaining decisions and knowing when to request human review. As AI systems become more able to act on their own, public agencies must clearly define who monitors them and who makes the final decision.
A mid-career guide to mastering AI
The video focuses on mastering AI at the mid-career level. It is aimed at professionals who already use AI daily and are ready to take on higher-stakes work.
Sources
- Hybrid Quantum Machine Learning Matches Classical Baselines but Clear Advantage Remains Elusive
- Systematic Review Details Enterprise Resource Planning Systems Transitioning Toward Artificial Intelligence
- Financial Times Names Cristina Criddle Chief Artificial Intelligence Correspondent
- Building safer AI products for children
- Persistence Model Outperforms Complex Machine Learning in Indonesian Rice Price Forecasting
- This school district wants AI in the classroom — with serious guardrails | The Metro
- The Agentic AI Revolution: How Manufacturers are Transforming Operations
- AI and Quantum: A One-Two Punch for Enterprise Security
- How to prepare the Pa for the responsible use of artificial intelligence
- Mastering AI at the Mid-Career Level
Comments
Please log in to post a comment.