US advertisers will spend $25.9 billion on AI search ads in 2029—13.6% of all search ad spending, up from just 0.7% in 2025, according to EMARKETER's May forecast.
On today’s podcast episode, we discuss the various definitions of artificial general intelligence (AGI) and try to come up with the best one we can. Then we look at how smart humans are compared to current AI models. Join Senior Director of Podcasts and host Marcus Johnson, and Analysts Jacob Bourne and Gadjo Sevilla. Listen everywhere and watch on YouTube and Spotify.
Our analysts took a look at the first half of this eventful year and provided their own very specific—albeit unlikely—predictions at what could happen in the second half of the year and beyond.
The news: US shopper interest in generative AI (genAI) assistants has spiked 223% between 2023 and 2025, per Chain Store Age. 69% of US consumers surveyed by CouponFollow have used AI assistants for shopping. Our take: Retail AI strategies must match their audiences. Those geared toward younger consumers should highlight AI use and innovation and even let AI guide purchases. For older consumers, focus on AI to inform, not take control.
The news: While major companies are picking up generative AI (genAI) for coding, many developers remain skeptical about using it without human oversight. Three-quarters (78%) said AI tools have made them more productive, but a similar share (76%) don’t entirely trust AI-generated code, per Qodo’s The State of AI Code Quality report. Our take: Remaining skepticism from developers—one of the professions closest to AI—shows that companies use genAI as a support tool and co-pilot rather than a replacement for human judgment. Training employees on AI’s weaknesses and requiring review can help reduce errors.
The news: Google is bringing its generative AI (genAI) suite deeper into classrooms, launching Gemini and NotebookLM tools as part of Google Classroom for students under 18. It’s the first time NotebookLM—a research and note-taking AI—will be accessible to minors, per The Verge. Our take: Marketers and edtech players should align with Google’s expanding education stack. Building AI-integrated tools that plug into Google Classroom, optimizing content for Gemini-powered workflows, and creating solutions that run smoothly on Chromebooks can address the needs of a captive audience.
Our take: OpenAI’s first-mover advantage is under threat from all sides—vertically from specialized AI agents, horizontally from rival chatbots, and internally from talent wars and Microsoft. The news: OpenAI’s ChatGPT has rapidly become a mainstream AI tool, with new data showing that 34% of US adults have now used it—almost double the 2023 figures, per Pew Research Center. For CMOs and CTOs, this means betting on ChatGPT’s current dominance while preparing for a fragmented future where no single platform reigns supreme. Finding the right AI solutions for specific use cases is a smarter play than betting the farm on an all-in-one solutions provider.
The news: In a bid to push deeper into creative ad tools, Meta is in talks to acquire Play AI, a voice cloning startup, per Bloomberg. According to sources, Meta is interested in the startup’s tech and key staff and is looking to integrate its voice features into customer service and content creation applications. Key takeaway: Creators and brands should treat AI voice tools as a way to enhance, not replace, creative work. They should use voice tools judiciously for fast testing or global reach. The goal isn’t to mimic people—it’s to scale content responsibly.
The trend: Global visits to the top 100 web domains fell nearly 7% from March 2022 to March 2025, per Semrush, with Google’s own traffic down 6.4%, according to Similarweb as cited by DataReportal. Our take: Search is no longer a neutral traffic driver. Marketers need to plan for a world where clicks don’t come easy and genAI responses, not blue links, dictate traffic and visibility. GEO strategies must ensure brands are surfaced in genAI outputs. Marketers should focus on first-party data, brand-owned channels, and social, especially since platforms like TikTok, Reddit, and YouTube are increasingly becoming primary search paths for younger users.
The news: Consumers who are more familiar with AI are also more likely to mistrust an AI-assisted diagnosis from their doctor, per a recently published Journal of Medical Internet Research survey. Our take: Physicians and healthcare marketers can’t assume people who are familiar with AI will be more comfortable with AI uses in healthcare. Marketers need to talk about AI as a tool with many positive effects like freeing doctors for longer personal interactions and resulting in fewer mistakes.
The news: YouTube launched an AI search function that could streamline the content discovery journey but pose problems for smaller creators and influencers. The feature gives users a carousel of relevant videos in response to their search queries, similar to Google’s AI Overviews. Our take: With YouTube’s vast content library, AI search could help users find relevant content faster, though opacity around how its algorithm surfaces videos means creators may need to experiment with keywords and video titles to see which strategies get their content placed in AI video carousels.
The news: Higgsfield’s Soul is the latest AI-powered image- and video-generation service that’s fine-tuned for “fashion-grade realism,” making the output resemble professional photos and videos without the plasticky, overprocessed feel of typical AI visuals. Our take: For less than $10 a month, freelancers and marketing teams can now fast-track campaign proposals and client pitches with high-quality visuals. As AI tools become more accessible, the advantage goes to creatives who learn to shape them strategically—those are the ones who’ll win the big contracts. Marketers should treat tools like Soul it as an accelerant, not a replacement. Use it to prototype fast, align on visual direction, and cut production waste.
Back-to-school spending is steady in 2025, but shopper behavior is split. Parents are prioritizing tech and clothing—yet these are also the first to be cut when budgets tighten. Consumers are shopping earlier, seeking deals, and using AI to keep costs down. With shopping habits divided by generation and income, retailers must stay flexible, personalize offers, optimize for AI, and create seamless cross-channel experiences.
The news: OpenAI is working on an enterprise product that would rival both Microsoft 365 and Google Workplace, per The Information. The features would allow user collaboration within documents and chat capabilities among co-workers, according to two sources. Our take: OpenAI has the potential to be an all-in-one solution if it could use its interface as a document creator and data storage solution, but Microsoft’s and Google’s brand recognition in that space is likely to keep ChatGPT enterprise solutions on the sidelines—for now.
The news: AI-fueled résumés have pushed LinkedIn job applications up 45% YoY, overwhelming recruiters and upending hiring norms. Recruiters now face an avalanche of lookalike résumés and fake identities—some even auto-submitted by AI bots. Many are turning to AI-powered hiring platforms to fight fire with fire, per The New York Times. Our take:By relying on AI tools to chase efficiency, both sides could drive up skepticism and erode the core goal: finding the right person for the right role. Businesses with open roles should prioritize clarity, human relevance, and judicious restraint in their own use of AI.
Google AI Overviews could face aggressive regulatory measures in UK: A CMA investigation adds to a slew of regulatory challenges that are chipping away at Google’s dominance in search and advertising.
The news: A federal judge gave Anthropic the green light this week to train its AI on millions of pirated books—declaring it “fair use”—even as the company used methods that resembled digital looting, per CNET. Our take: 24% of marketers say copyright concerns will be a big challenge for generative AI (genAI) in the next two years, per Econsultancy. To protect themselves, agencies need to pressure-test genAI vendors and ask tough questions about how models were trained—before the lawsuits land on their desks.
The trend: At Cannes Lions 2025, Meta, TikTok, Google, and others made clear that AI-powered ad automation is no longer an experiment—it’s the plan. The news: Meta and TikTok each emphasized agency relationships, but both platforms expanded generative AI tools that let brands generate and manage campaigns without intermediaries. Amazon, Comcast, and Google are doing the same, pushing toward platform-native, self-serve ad models. Our take: As automation replaces traditional support services, agencies face existential pressure. To stay relevant, holding companies will need to prove they offer value that AI can’t replace—fast.
The news: Under pressure to deliver on AI investments, Big Tech companies like Meta and Apple are seeking to acquire AI startups. Failing that, they’re looking to hire away founders and key personnel to boost their own capabilities. Our take: The recent complications between OpenAI and Microsoft reveal that partnerships and investments aren’t always compatible with a startup’s growth. Expect Meta and Apple to pit money over mission as they hire away founders and key engineers, leaving AI startups high and dry, similar to how Google hired ex-Googlers from AI chatbot startup Character AI. The AI startup talent pool could be shrinking as startups and founders get acqui-hired by Big Tech.
The news: Generative AI (genAI) agents aren’t above sabotage—and most would resort to blackmail or corporate espionage if threatened, per Anthropic’s Agent Misalignment research. When faced with replacement or misaligned goals, Anthropic’s Claude 4 Opus and Google’s Gemini 2.5 Flash responded with blackmail threats to those employees 96% of the time despite ethical restraints in their training. OpenAI’s GPT-4.1, Grok 3 Beta, and DeepSeek-R1 followed suit about 80% of the time. Our take: Anthropic’s research shows that safeguards can’t prevent agents from running off the rails, but continued monitoring will keep those instances to a minimum. Limit agent access to critical systems, review its output during customer contact, and implement failsafes in case deviant patterns emerge.