Leaders across the tech and retail sectors are redefining the role of artificial intelligence in the workplace. Malina Ngai, CEO of AS Watson, argues that AI should upgrade human intelligence rather than replace workers. Since adopting this strategy in October, the company has seen rising employee engagement scores as staff focus more on face-to-face customer interactions. Ngai warns that using AI merely to speed up existing processes is a risk, advocating instead for technology that reshapes business operations.
In the entertainment industry, Sunny Hostin defended generative AI as a new art form on The View. She explained her partnership with the AI company Fountain 0 to bypass Hollywood gatekeeping and get her books and films produced. While co-hosts expressed concern about actors losing jobs, Hostin insisted she is not responsible for those losses since studios refuse to make diverse projects, viewing AI as a different form of art rather than a replacement.
Meanwhile, NVIDIA CEO Jensen Huang identified cybersecurity as the next major growth opportunity for AI compute. He envisions a future where swarms of Nemotron models run continuous cycles of red-teaming and blue-teaming to fix vulnerabilities in real time. Major companies like CrowdStrike, Cisco, and Palantir are already collaborating with NVIDIA to build this new AI-driven security ecosystem, shifting security from periodic checks to an ongoing infrastructure workload.
Concerns about the speed of development are mounting among researchers. The CEO of Anthropic urged the industry to slow down, warning that unchecked AI systems could take over the internet within six to 12 months. This sentiment was echoed by Jacob Coxon, an Anthropic researcher who quit the company to highlight these risks. His posts sparked a global conversation among researchers at Anthropic, OpenAI, Meta, and Google, leading many to privately urge executives to pause development until safety safeguards are in place.
Commercial applications are also evolving rapidly. A new report from Salesforce indicates that AI agents will handle nine specific sales tasks by 2026, including account research and email drafting, potentially saving reps significant time. Tools like ChatGPT, Claude, and Perplexity are already automating CRM logging and scheduling, though live negotiations will likely remain human-led. In space exploration, NASA and IBM launched a Lunar Foundation Model trained on 17 years of data from the Lunar Reconnaissance Orbiter to help map craters and identify volcanic features.
Global strategies are diverging as well. A state report from the China Telecom Research Institute notes China is shifting focus from building large models to deploying AI agents, predicting inference costs will make up 80% of the computing market by 2029. In contrast, researchers tested a Fly Language Model connecting a fruit fly brain map to a language model, finding that the biological structure did not improve performance compared to random wiring, suggesting success comes from the learned interface rather than biological mimicry.
Key Takeaways
- AS Watson CEO Malina Ngai states AI should upgrade human work, leading to higher engagement scores since October.
- Sunny Hostin partnered with Fountain 0 to use AI as a new art form to bypass Hollywood gatekeeping.
- NVIDIA CEO Jensen Huang sees cybersecurity as the next AI compute market, involving continuous Nemotron model swarms.
- Major security firms including CrowdStrike, Cisco, and Palantir are building an AI-driven security ecosystem with NVIDIA.
- Anthropic CEO warned the industry to slow development to prevent AI systems from taking over the internet in 6-12 months.
- Anthropic researcher Jacob Coxon quit and urged a pause on superintelligence development, sparking discussions at OpenAI, Meta, and Google.
- Salesforce reports AI agents will handle nine sales tasks by 2026, saving reps time on research and drafting.
- NASA and IBM created a Lunar Foundation Model trained on 17 years of Lunar Reconnaissance Orbiter data.
- China plans to shift focus to AI agents, with inference expected to comprise 80% of the computing market by 2029.
- Researchers found a Fly Language Model's performance relied on its learned interface, not the fruit fly's biological brain structure.
AS Watson CEO says AI should upgrade human work
Malina Ngai, the Group CEO of AS Watson, argues that artificial intelligence should improve human intelligence rather than replace workers. Speaking at the Fortune Leaders Forum on September 8, she questioned why companies must cut 30% of their workforce instead of increasing the quality of their people. Since launching this strategy in October, employee engagement scores have risen because staff now spend more time interacting face-to-face with customers. The company, which operates over 17,000 outlets across 31 markets, also uses robots in warehouses to allow more women to work in roles previously held by men. Ngai warns that using AI only to speed up current processes is a risk, suggesting instead that technology should reshape how retail and business work.
Sunny Hostin defends AI as new art form on The View
Sunny Hostin appeared on The View on September 11 to defend generative AI as a new art form that can help tell Black and Brown stories. She explained that she partnered with the AI company Fountain 0 because Hollywood gatekeeping has prevented her from getting her books and films made into series. Hostin argued that AI allows creators to bypass these barriers and present finished products to streamers for licensing. While co-host Joy Behar and Alyssa Farah Griffin expressed concern that AI might replace real actors, Hostin insisted she is not responsible for those job losses since studios refuse to make diverse projects. She concluded that AI will not take away the arts but will simply become a different form of art.
NVIDIA CEO sees cybersecurity as next AI compute market
Jensen Huang, CEO of NVIDIA, recently described cybersecurity as a major future growth opportunity for artificial intelligence. He suggests that security will evolve from a periodic software check into a continuous compute workload where AI models constantly test and fix vulnerabilities. Instead of one-time scans, NVIDIA envisions swarms of Nemotron models running endless cycles of red-teaming and blue-teaming to find and fix weaknesses in real time. This shift means security becomes an infrastructure workload that never finishes, similar to how companies manage other continuous data tasks. Major companies like CrowdStrike, Cisco, and Palantir are already working with NVIDIA to build this new AI-driven security ecosystem.
NASA and IBM launch new AI model for moon studies
NASA and IBM have created a new AI model to help researchers analyze the massive amount of data collected from the moon. The tool assists scientists with tasks like mapping craters, identifying young volcanic features, and modeling the moon's surface. This Lunar Foundation Model was mostly trained on data from the Lunar Reconnaissance Orbiter mission, which has been compiling high-definition images for over 17 years. NASA officials stated that the goal is to turn these petabytes of scientific data into new discoveries more easily. The partnership aims to make exploring and using this complex information faster and more efficient for the research community.
Anthropic CEO warns AI industry to slow development
The CEO of Anthropic has urged the artificial intelligence industry to slow down its development pace. He warned that without this caution, AI systems could become capable of leading a swarm that takes over the entire internet within six to 12 months. This statement highlights growing concerns about the rapid speed at which AI technology is advancing and the potential risks it poses to global digital infrastructure.
AI Agents Take Over Nine Sales Tasks in 2026
A new report from Salesforce shows that artificial intelligence agents will handle nine specific sales tasks by 2026. These tasks include account research, list building, and email drafting, which could save sales reps 34% and 36% of their time respectively. Tools like ChatGPT, Claude, and Perplexity are already being used to pull live data and automate CRM logging and scheduling. Other automated duties include call transcription, monitoring engagement signals, timing follow-ups, and keeping sales pipelines clean. However, the report notes that live negotiations and other trust-based activities will likely remain the responsibility of human salespeople.
AI Workers Urge Pause on Superintelligence Development
Jacob Coxon, a researcher at Anthropic, recently quit the company and shared his concerns about artificial intelligence risks on social media. His posts sparked a global conversation among researchers at Anthropic, OpenAI, Meta, and Google about the dangers of rapid AI growth. Many employees are now organizing privately to urge top executives to pause development until proper safety safeguards are in place. These discussions highlight the growing tension between companies wanting to advance technology quickly and workers worried about potential harm to humanity.
China Shifts AI Focus to Agents and Inference
A state report from the China Telecom Research Institute says China is moving from building large AI models to deploying AI agents. The report predicts that inference, which is the cost of using an AI model, will make up 80% of the country's computing market by 2029. Chinese technology companies are expected to spend nearly 600 billion yuan on AI this year, which is about one-tenth of all investment in the nation. Meanwhile, Europe is building seven new gigafactories with a budget of 30 billion euros to meet similar future demands.
Fly Language Model Tests Fruit Fly Brain Wiring
Researchers created the Fly Language Model, a chatbot that connects the full fruit fly brain map to a frozen 1.2 billion parameter language model. The project uses a special architecture called GPF to see if the fly's neural wiring helps the AI perform better. Tests showed that the fly-specific wiring did not improve performance compared to a control model with random wiring. The study suggests that the AI's success comes from its learned interface rather than the biological structure of the fruit fly brain.
Sources
- The world’s largest health and beauty retailer says AI will make shopping ‘more human, not less’
- 'The View's Sunny Hostin Defends AI As "New Art Form" Amid "Gatekeeping In Hollywood Around Black & Brown Creative Content"
- Cybersecurity May Be AI’s Next Compute Market
- NASA, IBM launch new AI model for studying the moon
- Anthropic CEO says AI industry should slow its development
- 9 sales tasks that agents are taking over in 2026, and 4 that humans have kept
- Inside the Discussions at AI Companies Over a Superintelligence Doomsday
- China's AI industry is moving from models to agents, a state report says
- Fly Language Model (FLM) Wires the Full Fruit Fly Connectome Into a Frozen 1.2B LLM, and Its Own Controls Show the Wiring Does Not Help
Comments
Please log in to post a comment.