By Haddon Libby
In the waning months of 2025, the artificial intelligence sector underwent what many experts describe as its most intense period of progress to date. Between November and December, leading labs released a flurry of frontier models that dramatically advanced reasoning, multimodality, and agentic capabilities, pushing AI closer to human-expert performance across complex tasks.
xAI, founded by Elon Musk, rolled out Grok 4.1 on November 17, emphasizing real-time data integration from X and superior performance in mathematical and coding benchmarks. Google followed with Gemini 3 in mid-November, including the efficient Gemini 3 Flash variant in December. This series set new records in multimodal reasoning (handling text, images, video, and code) with standout results on challenges like GPQA Diamond and Humanity’s Last Exam (HLE). These have emerged as two of the toughest current benchmarks; they push beyond older tests by demanding deep, expert-level reasoning that is difficult to fake with memorization or simple search.
Anthropic launched Claude Opus 4.5 around November 24, dominating in agentic workflows and autonomous coding. It achieved over 80% success rates on the SWE-Bench software engineering benchmark. OpenAI, responding to competitive pressure, accelerated its timeline to release GPT 5.2 in December, building on earlier iterations with enhanced adaptive reasoning, tool use, and vision capabilities.
This rapid succession marks what industry observers called a “vertical acceleration.” AI models now routinely surpass 90% on once-daunting PhD-level science and math tests. Leaders including Sam Altman, Demis Hassabis, and Dario Amodei hailed these shifts as pivotal steps toward more autonomous, reliable AI systems capable of self-improvement and real-world problem-solving.
However, these technical triumphs carry profound societal implications. On employment, productivity gains have been uneven. Industries embracing technology report faster wage growth and a high demand for AI-fluent skills. Software development, for instance, could see robust growth as AI handles routine coding, freeing humans for creative oversight. Yet displacement is accelerating in entry-level white-collar positions. Amodei warned in 2025 that up to half of such jobs could vanish within five years, potentially driving unemployment to 10-20% in a short window. Tech giant layoffs, often framed as “efficiency” measures, already number in the tens of thousands. Without aggressive reskilling and policy support, economists fear significant short-term economic strains.
Ethically, the pace has outstripped safeguards. Biases persist in training data, risking amplified discrimination in hiring, lending, and justice systems. Deepfakes and misinformation have surged, influencing elections and personal harms, while AI companions have raised alarms over mental health impacts among vulnerable users. These incidents have prompted calls for stronger governance, such as the EU AI Act. At global forums, leaders have stressed human-centric approaches to prevent the erosion of trust, privacy, and democratic norms.
Amid the risks lies transformative promise in human augmentation. AI increasingly acts as a collaborator, amplifying creativity and expertise. Agentic systems manage workflows, enabling workers to oversee digital “fleets” rather than perform rote tasks. This symbiosis could boost productivity dramatically, with markets for augmentation technologies projected to reach trillions. Advocates envision expanded discovery in medicine, education, and science, where humans focus on empathy and innovation while AI handles scale and iteration.
As 2026 unfolds, the late-2025 surge underscores a critical juncture. The race among OpenAI, Google, Anthropic, and xAI has delivered capabilities once thought distant, but it also amplifies questions of equity and accountability. Harnessing this power thoughtfully could usher in an era of unprecedented human flourishing. The path forward depends not just on code, but on collective choices about the society AI will help build.
Haddon Libby is the Founder and Chief Investment Officer at Winslow Drake Investment Management, a locally based RIA firm. For more information on our services, please visit www.WinslowDrake.com.
