AI in Marketing

The New York Times recently published a quiz asking readers to distinguish AI-generated prose from passages by published authors. More than 86,000 people participated. The results: 54 percent preferred the AI versions, and most struggled to identify which was which.

On this week's episode of The Artificial Intelligence Show, Paul Roetzer and Mike Kaput argued the quiz asks the wrong question. The more useful question isn't whether AI can write well. It's when humans should use it.

Roetzer, who has written three books, proposed a framework with five levels. At level zero, the human writes everything. This applies when voice matters most: essays, speeches, personal letters. At level four, AI writes autonomously with minimal oversight. This suits high-volume, low-personality content like product descriptions.

Most business writing falls somewhere in between. The deciding factor is authenticity. When audiences expect you, you need to show up. When the value is information rather than self-expression, AI can take on more.

Kaput acknowledged the emotional weight of these findings for writers. But he cautioned against denial. "If you're offended because you think AI can't do these things, you need to update your priors," he said.

Roetzer added that writing remains central to clear thinking. As AI handles more routine content, organizations will need to reconsider how they train people to think critically in the first place.

Next.

The BBC reports that US AI firm Anthropic is hiring a chemical weapons and explosives expert to prevent misuse of its software, fearing it could be used to create weapons. The role requires at least five years of experience in chemical weapons defense and knowledge of radiological dispersal devices. This move mirrors a similar position advertised by OpenAI, which offers a higher salary for a researcher in biological and chemical risks. Some experts express concern about AI systems handling sensitive weapons information, noting the lack of international regulation. Anthropic is also taking legal action against the US Department of Defense, which labeled it a supply chain risk, while OpenAI has negotiated its own contract with the US government.

On a different note.

Mistral, a French AI startup, has launched Mistral Forge, a platform enabling enterprises to create custom AI models using their own data, as announced at Nvidia's GTC conference. Unlike competitors OpenAI and Anthropic, which focus on consumer markets, Mistral targets enterprise clients, aiming to surpass $1 billion in annual recurring revenue. Mistral Forge allows companies to train AI models from scratch, offering greater control over data and model behavior, potentially improving performance with non-English or domain-specific data. This approach contrasts with existing methods that adapt models at runtime using company data. Mistral's platform includes a library of open-weight AI models, such as Mistral Small 4, and provides guidance on model and infrastructure choices, though final decisions rest with the customer. Mistral's team of forward-deployed engineers assists clients in optimizing data usage and model evaluation. Early adopters of Forge include Ericsson, the European Space Agency, and ASML, reflecting its appeal to governments, financial institutions, manufacturers, and tech companies. These organizations benefit from tailored models that meet specific language, compliance, and customization needs, reducing reliance on third-party providers and mitigating risks of model changes or deprecation.

Meanwhile.

Nvidia CEO Jensen Huang discussed the evolution of accelerated computing and Nvidia's strategic direction in a recent interview following his GTC 2026 keynote in San Jose. Huang reflected on Nvidia's history, emphasizing the significance of the programmable shader and CUDA, which have been pivotal in Nvidia's growth across various industries. He highlighted the company's expansion into new sectors and the importance of accelerating software for AI applications, noting that AI will increasingly utilize tools originally designed for humans. Huang addressed Nvidia's burgeoning CPU business and the acquisition of Groq, asserting that these moves do not represent a shift in strategy but rather an extension of Nvidia's capabilities. He also discussed the challenges of scarcity in the AI stack and the geopolitical implications of Nvidia's operations in China. Huang expressed frustration with the influence of pessimistic voices in Washington. The conversation touched on Nvidia's role in building AI infrastructure globally, emphasizing the need for integrated systems to optimize performance and investment. Huang underscored the necessity of developing new models and architectures to advance AI applications beyond language models, highlighting innovations like the hybrid architecture of transformers with SSM.

In other news.

OpenAI has entered a contract to provide its AI models to U.S. defense and government agencies through Amazon's cloud unit, marking a shift from its previous focus on unclassified government use. This deal follows the Pentagon's decision to drop Anthropic, a former key AI supplier, due to its refusal to allow unrestricted military use of its AI for domestic surveillance and autonomous weapons. The Pentagon labeled Anthropic a "supply chain risk," severing its government ties. OpenAI's partnership with Amazon Web Services positions it to support classified operations, highlighting the competitive landscape for government and defense contracts. This move could also attract large corporate clients, as public sector work is often seen as a sign of trust and reliability.

Shifting gears.

Over 45,000 technology jobs have been cut globally in early 2026 as companies restructure around artificial intelligence, marking a significant shift in the industry. Amazon leads with 16,000 layoffs in January, part of a larger reduction since October 2025. Beth Galetti, Amazon's senior vice president, stated the move aims to streamline operations. Block, the fintech firm, announced a 40% workforce reduction, citing AI's role in enhancing efficiency. CEO Jack Dorsey emphasized the impact of AI tools on company operations. Dell Technologies reported a 10% workforce decline, continuing a trend of annual reductions. Atlassian plans to cut 1,600 jobs to focus on AI development. Despite these cuts, companies like Amazon and Block report strong financial performance, raising questions about AI's role in job reductions. Analysts suggest AI is reshaping labor, with potential tech-sector layoffs reaching 265,000 by year-end. The broader impact extends beyond tech, with U.S. planned layoffs in January reaching their highest since 2009.

Elsewhere.

Eight major tech companies, including Google, Microsoft, Meta, and Amazon, have signed the "Online Services Accord Against Scams" to combat online fraud, as global scam losses surpassed $1 trillion. This agreement, announced at the UN Global Fraud Summit in Vienna, aims to formalize the sharing of threat intelligence and best practices among the signatories. The Global Anti-Scam Alliance (GASA) reported that US consumer losses reached $16.6 billion in 2024, with cyber-enabled crime accounting for 83% of these damages. The FBI noted a significant rise in complaints, with 859,532 cases reported in 2024, marking a 33% increase from the previous year. The accord, which carries no penalties for non-compliance, seeks to enhance user safety features and improve communication between companies to disrupt scam networks more effectively. The agreement follows previous voluntary pledges, such as the UK's 2023 Online Fraud Charter and the 2024 AI Election Accord. Despite these efforts, the gap between detection and enforcement remains a challenge, with only 0.05% of cybercrimes resulting in prosecution. The UN Summit, supported by GASA, Meta, and Santander, continues to address these global fraud issues.

Finally.

Burger King is undergoing a significant brand overhaul, acknowledging past mistakes and actively seeking customer feedback to guide its transformation. The fast-food chain has launched a $700 million initiative to improve its image after losing its position as the second-largest US burger chain in 2020. Changes include updating restaurant operations, technology, and menu items, as well as eliminating the king mascot, which many found unsettling. Burger King President Tom Curtis has encouraged direct customer interaction by sharing his phone number on social media, receiving over 20,000 responses. The feedback has been instrumental in identifying areas for improvement, such as packaging changes to prevent burger flattening. The brand's recent ad campaign, reminiscent of Domino’s 2009 “Pizza Turnaround,” candidly addresses its shortcomings, aiming to rebuild trust and satisfaction. Burger King has climbed to sixth place in overall satisfaction among top US quick-service restaurant brands, up from tenth in 2020. The company remains committed to maintaining open communication with customers, ensuring the brand reset is perceived as genuine and ongoing. This approach reflects a strategic shift towards a more customer-centric model, with continuous feedback shaping future developments.