Skip to main content

Luxury brands are defined by exclusivity, artisan craftsmanship, and premium pricing. Gucci’s recent use of AI-generated images for its Primavera Fashion Show advertising, however, has sparked criticism. The AI ads featured surreal, computer-created images that disappointed many fashion enthusiasts who expect traditional, high-quality craftsmanship in every aspect of the brand—including its marketing.

While AI can be cost-effective compared to traditional photo shoots, many consumers feel it undermines Gucci’s luxury identity. Online reactions ranged from disappointment to outright rejection, suggesting that luxury consumers expect a higher standard that AI-generated content fails to meet. This backlash highlights the risk luxury brands face when trying to cut corners in their advertising efforts.

Why Gucci’s Use of AI in Advertising Misses the Mark for Luxury Brands

Artificial intelligence (AI) is transforming institutional work across sectors, but government agencies face unique challenges that consumer AI tools like ChatGPT or Copilot are ill-equipped to solve. Government operations are deeply siloed, stretched thin by budget limits and manpower shortages, and require complex coordination across multiple organizations. Unlike commercial AI designed for private sector processes, government AI must integrate seamlessly across various agencies, nonprofits, and partners to be truly effective.

Traditional chatbots lack the context and access to dispersed, sensitive government data needed to support public service tasks such as managing reimbursement processes or coordinating emergency responses. Embedded AI, however, lives within the platforms used daily by government workers and supports their specific workflows and decision-making. This deeper integration allows AI to not just provide information, but also guide next steps, follow up on tasks, and maintain continuity despite personnel changes.

By embedding AI in government operations—not adding extra tools—public servants can work faster, coordinate better, and serve the public more effectively. The ultimate aim is to improve government programs that impact daily lives, from nutrition assistance to veterans services, delivering better community outcomes through smarter technology.

Why Embedded AI is Essential for Modern Government Efficiency

Recent developments at Microsoft's Xbox division have taken the gaming community by surprise. Phil Spencer, who led Xbox for nearly 12 years, unexpectedly retired, and Asha Sharma, previously president of Microsoft's CoreAI product, has been appointed to lead Xbox. Despite her lack of a classic gaming background, Sharma faces significant challenges as Xbox's financial metrics show decline, with drops in hardware revenue, overall gaming revenue, and content services including Game Pass subscriptions. Her appointment also symbolizes a shift in focus as Microsoft balances between console sales and expanding its subscription services across platforms. Sharma's leadership style will be critical as Xbox navigates inventory challenges, changing consumer preferences, and competition, while continuing to leverage exclusive franchises like Call of Duty, Halo, and others. The company is poised to redefine its strategy in an evolving gaming market with an emphasis on both hardware and subscription models, requiring innovative thinking and steady guidance to reclaim lost ground.

Microsoft Appoints AI Leader to Oversee Xbox Amidst Industry Challenges

As AI becomes more ingrained in daily tasks, frustration with its quirks is on the rise. A report by Adobe Acrobat and Firefly reveals that 91% of surveyed AI users have abandoned AI-generated tasks due to these emotions. Crafting effective AI prompts can be challenging, with users often losing patience after several attempts—four tries for image generation and two to four for text tasks like emails or social media posts. Interestingly, men are 80% more likely than women to shout at AI in all caps, mistakenly thinking it may improve outcomes. Sectors like finance and education tend to encourage politeness, with frequent use of "please" when interacting with AI. Despite men showing 15% more confidence in their prompting skills, their success rate is only marginally better. Tips to reduce frustration include breaking tasks into steps, saving strong prompts, fact-checking, and providing clear examples. However, being polite or yelling at AI doesn’t affect performance.

Men Are 80% More Likely to Yell at AI in All Caps Than Women, Study Finds

OpenAI CEO Sam Altman has defended the energy-intensive processes behind artificial intelligence by comparing the resource demands of training AI models to the energy and food humans require over a lifetime. This perspective sparked significant backlash, especially from climate experts who argue that the comparison misses key points about the environmental impact and resource consumption unique to AI. Experts emphasize that humans and AI operate differently in terms of energy use, noting that Altman’s view overlooks the ecological costs embedded in creating the data that trains AI models. Critics also highlight the growing strain AI places on energy and water resources, pointing to projections that AI-driven demand on water and electricity will surge significantly in the coming decades. Altman acknowledges energy concerns and calls for a faster transition to renewable sources like nuclear, wind, and solar. However, climate scientists caution that this defensive stance echoes problematic tech world beliefs such as longtermism and techno-utopianism, which may underplay the immediate risks to the environment, including the climate crisis.

Sam Altman Defends AI's Energy Use Amid Criticism From Climate Experts

Welcome to AI Decoded, Fast Company’s weekly newsletter that distills the most crucial updates in artificial intelligence. Sign up to get this newsletter delivered every week via email.

Despite AI’s promise to transform industries, its widespread impact remains elusive. Major companies tout AI’s potential to revolutionize business, but real-world outcomes vary widely. Organizations often lack the culture, strategy, or data maturity to fully integrate AI, leading many to remain stuck in pilot programs without clear returns. A 2025 MIT study found that 95% of large enterprises saw no measurable profit and loss impact from their substantial investments in generative AI, though more recent research from Wharton offers a cautiously optimistic outlook, with most leaders reporting gains and planning to boost AI budgets.

The uneven pace of adoption means some sectors, like software and retail, are seeing tangible efficiency improvements, while others struggle. Additionally, initial implementation phases require significant human effort to train and correct AI, which can dampen productivity gains. Ultimately, the key obstacle to AI’s success isn’t the technology itself but the time and cultural shifts needed within organizations to realize its benefits.

Meanwhile, new benchmarks indicate AI still falls short on complex, real-world gig tasks, underlining the gap between hype and practical capability. Ethical tensions also arise, exemplified by Anthropic’s refusal to let its AI be used for certain defense applications, highlighting a broader debate on AI’s responsible use and industry-government relations.

For deeper insights and updates on AI, business innovation, and technology trends, consider Fast Company Premium.

Why AI’s Real Challenge Lies in Adoption, Not Intelligence

In the fast-evolving AI industry, innovations are outpaced in weeks rather than months or years. Anthropic's recent release of Opus 4.6 showcased leading-edge AI capabilities but was shortly challenged by China's Z.ai with their GLM-5 model. Despite no direct copying allegations, the rapid deployment and subsequent modification of these models highlight a trend: the window for exclusive AI breakthroughs is shrinking. Industry experts note that improvements increasingly come through ongoing tweaks like reinforcement learning and extended context capabilities rather than initial training. This results in swift release cycles akin to software updates. American companies have raised concerns about competitive practices such as distillation, where extensive model interactions extract insights to replicate core AI performance. Licensing terms add complexity, often restricting 'open-source' use based on deployment scale, creating ambiguity fueling competitive strategy shifts. The market is anticipated to split between accessible, self-hosted AI variants for common uses and high-end systems for complex applications. Experts suggest that efforts to completely block knowledge extraction from new models may be futile, as replication attempts appear inevitable upon release.

AI Model Advancements and Rapid Imitation Narrow Competitive Edge

Sebastian Siemiatkowski, CEO of Klarna, the Swedish fintech company, announced a significant reduction in staff over the coming years. The current workforce of about 3,000 is down from 7,000 four years ago, with expectations to drop below 2,000 by 2030—a decrease influenced by layoffs, attrition, and AI-related efficiencies. Despite buy now, pay later services gaining popularity, the integration of AI means fewer employees are needed. Siemiatkowski aligns with Anthropic CEO Dario Amodei’s concerns about AI's profound effects on jobs, forecasting substantial short-term upheaval but adopting AI tools within Klarna to handle more tasks. This move initially led to customer service challenges, prompting a return to a balanced approach with human agents tackling complex cases.

Klarna CEO Predicts Workforce Will Shrink by 1,000 by 2030, Citing AI Impact

OpenAI, Google, and Perplexity are close to securing approval to sell their AI technologies directly to the U.S. government via their own cloud platforms, according to an insider. This approval, expected to be on a "low impact" pilot basis, represents a significant move toward operational independence. Historically, AI firms have depended on established government contractors like Microsoft, Palantir, and AWS to host their services for federal use, which limited their control over deployment. In contrast, Anthropic's reliance on partners like Palantir has sparked complications, exemplified by a Pentagon dispute over AI use in military contexts and concerns about autonomous weapons. Seeking autonomy, OpenAI, Perplexity, and Google pursued accelerated FedRAMP 20x security reviews last year and appear near approval, positioning themselves to engage the government directly. Anthropic has not taken the same path, though it acknowledges the benefits of direct government service offering.

OpenAI, Google, and Perplexity Preparing for Direct Government Cloud Hosting Approval

Variant, a generative design platform, has introduced an AI-powered eyedropper tool that enables designers to capture the "vibe" of one AI-generated interface—typography, spatial layout, and colors—and transfer it to another. This tool mimics traditional design methods, offering direct manipulation rather than relying solely on text prompts, which often feel imprecise. Although the underlying AI generates somewhat uniform and flat designs, the eyedropper’s animation and interactive style transfer provide an innovative and engaging user experience. This approach builds on decades of digital design history, from SuperPaint’s color sampling to Adobe Illustrator’s style stealing, highlighting a need for more precise, visually guided AI tools over ambiguous text commands. The tool advocates for design workflows that mirror human visual intuition, urging generative AI developers to prioritize direct manipulation in their interfaces.

Variant’s AI Eyedropper Brings Classic Design Precision to Generative UI