In a recent LinkedIn post, Kieran Flanagan discusses a significant and somewhat unsettling shift he’s observed in the latest AI models, particularly Opus 4.6 and GPT-5.3 Codex. Flanagan, a keen observer of AI advancements, highlights a newfound capability in these models that goes beyond technical correctness, suggesting they are developing a form of ‘judgment’ and ‘taste’.
Flanagan’s observations are prompted by a noticeable change in AI performance over the past two weeks. He points to a sentiment echoed by others in the AI field, citing Matt Schumer’s post, which described the experience as AI having:
“It had something that felt, for the first time, like judgment. Like taste. The inexplicable sense of knowing what the right call is that people always said AI would never have.”
This development challenges Flanagan’s previous perspective. He admits that in creative tasks, such as brainstorming content or exploring different hooks and angles, the AI’s output was previously only ‘average,’ which he found reassuring as it affirmed his own expertise gained through years of experience. However, this has recently changed.
The Emergence of AI ‘Taste’
Flanagan recounts a personal experience where he was collaborating with Opus 4.6 on a creative task late at night. He found that the AI began making choices that, in some instances, surpassed his own. This led to a moment of reflection, described as feeling ‘a little sad,’ akin to how developers might feel when AI assists in writing code.
According to Flanagan, this evolution is particularly significant because OpenAI has stated that GPT-5.3 Codex played a role in its own creation, debugging its training, managing deployment, and diagnosing evaluations. He emphasizes the recursive nature of this self-improvement:
“The latest models have taste and are helping themselves grow. Recursive loops like this don’t slow down; they only speed up.”
This observation aligns with a broader feeling within the AI community that something fundamental has shifted. Flanagan notes that while many were aware of this change, it gained wider recognition and validation after Matt Schumer articulated it, resonating with thousands of others.
Navigating the Unsettling Shift: Opportunity Amidst Change
Despite the unsettling nature of such rapid advancement, Flanagan maintains an optimistic outlook, viewing it as a unique opportunity. He argues that while the AI models are becoming more capable assistants, human domain expertise remains critical for directing their application and framing their value. As Flanagan states:
“My first line in this post used the word ‘unsettling’. I used it purposely. All change is unsettling, and I believe it’s the job of those in the weeds to help others prepare for this shift.”
He believes that this period presents a ‘once-in-a-lifetime opportunity’ for individuals to develop and leverage the skills that will matter most as the nature of work transforms. Flanagan underscores the importance of having the original idea and providing clear direction, even when AI produces incredible creative output. His advice for navigating this evolving landscape is to focus on controllable aspects, learn continuously, and remain curious.
Flanagan encourages a proactive approach, suggesting that understanding and adapting to these changes is key to seizing the new opportunities emerging in the AI-driven future of work.
📝 About This Content
This article is based on insights shared by Kieran Flanagan on LinkedIn.
📅 Originally posted on February 11, 2026 | View original post on LinkedIn →