## Sources 1. [TBM 413: In That Space Is Our Power](https://cutlefish.substack.com/p/tbm-413-in-that-space-is-our-power) 2. [AI Is Here, But The Hard Parts Haven't Changed](https://joereis.substack.com/p/ai-is-here-but-the-hard-parts-havent) 3. [The data reliability question you're avoiding](https://andrewrjones.substack.com/p/the-data-reliability-question-youre) 4. [The Semantics of Semantics](https://jessicatalisman.substack.com/p/the-semantics-of-semantics) 5. [TBM 412: Institutionalized Overload (Now With AI)](https://cutlefish.substack.com/p/tbm-412-institutionalized-overload) 6. [I’m at SFO waiting for my flight - Ask me anything](https://joereis.substack.com/p/im-at-sfo-waiting-for-my-flight-ask) --- ### "AI Is Here, But The Hard Parts Haven't Changed" by Joe Reis * **Universal adoption of AI:** Nearly all data professionals are using AI tools in their daily workflows, with **Claude being the dominant choice at 49%**, significantly outpacing GitHub Copilot and ChatGPT [1-4]. * **Speed vs. Quality:** While **57% of respondents say AI helps them write code significantly faster**, there is growing concern that churning out code faster will lead to quality issues and production becoming a "cesspool" [3, 5]. * **The new technical debt:** AI is creating a new form of technical debt where engineers are deploying generated code and systems that **they do not fully understand or read entirely** [6]. * **The importance of fundamentals:** Despite the rise of AI, **49% of practitioners believe data modeling and semantic layers will be the most critical skills in 2027** [4, 7]. AI is highly capable of generating code, but it still struggles to understand the contextual meaning of an organization's data [8]. * **Enduring bottlenecks:** The core challenges in data engineering remain deeply human and organizational. The biggest bottlenecks are **legacy systems (25%), lack of leadership direction (21%), and poor requirements (19%)**—issues that AI cannot fix [9]. * **Career optimism:** Practitioners are largely optimistic, viewing AI as an empowering tool rather than a replacement. However, professionals who use AI to bypass learning the fundamentals risk becoming replaceable "button pushers" [10, 11]. ### "I’m at SFO waiting for my flight - Ask me anything" by Joe Reis * **Live Engagement:** This source serves as a placeholder for a live "Ask Me Anything" (AMA) video session recorded by the author while waiting for a flight at San Francisco International Airport [12]. * **Ongoing Topics:** The post points to previous AMAs that cover related industry topics, such as ontologies, data modeling, data engineering, and the future of AI [12]. ### "TBM 412: Institutionalized Overload (Now With AI)" by John Cutler * **The normalization of overload:** Modern workplaces have become so acclimated to endless work-in-progress and constant unplanned demands that **cognitive overload has been normalized and internalized as part of professional identity** [13, 14]. * **AI sustains the chaos:** Instead of using AI to fix broken organizational structures, **people are using these tools to navigate and sustain the existing cognitive overload** [13, 15]. AI allows workers to juggle more balls and process more context, artificially raising expectations further [15]. * **Illusion of disruption:** New technology rarely breaks an organizational paradigm on its own; it typically **reinforces the system it enters** [13]. Leaders push to use AI for efficiency but avoid changing the actual flow of power and decisions [13]. * **Resisting expansion:** In the current AI hype cycle, information and context will organically expand to fill available space. A crucial part of modern work is **actively resisting this maximalist expansion** rather than just adapting to it [16]. ### "TBM 413: In That Space Is Our Power" by John Cutler * **The burden of proof on change agents:** When advocating for better ways of working, change agents often face the **unfair burden of having to prove their perspectives are valid**, while dealing with immediate dismissal or invalidation from those comfortable with the status quo [17-19]. * **Protecting professional identity:** Both the change agent and the defender of the status quo are operating from vulnerability. **The need to be heard and the need to preserve the current system are both deeply tied to professional identity** [20, 21]. * **Perspective on workplace relationships:** It is vital to remember that workplace relationships are usually "situational friends" and should not be equated to deep, transcendent personal relationships like family [21, 22]. * **Goal clarification:** Change agents need to ask themselves if they genuinely want to change behavior—which often requires a subtle "show, don't tell" approach—or if their primary goal is merely to feel validated and listened to [22, 23]. * **Acknowledge your needs:** By deeply acknowledging one's own needs and separating them from the desire to control the situation, **a professional can prevent hurt feelings from escalating into unproductive conflict** [23, 24]. ### "The Semantics of Semantics" by Jessica Talisman * **A co-opted term:** The data industry's use of the term "semantic layer" is fundamentally flawed. It was originally **a retroactive marketing term coined in the 1990s by Business Objects** to describe a SQL generation tool, not a formal semantic system [25-28]. * **SQL compilation vs. Machine logic:** Modern BI semantic layers (like dbt, Looker, and AtScale) are essentially **SQL compilers and metric governance tools** that use declarative definitions to generate queries [29-31]. They do not perform automated reasoning, verify logical consistency, or use formal mathematics [29, 32]. * **The true meaning of semantics:** In computer science and the knowledge graph community, semantics refers to **formally defined, machine-interpretable meaning based on mathematical logic** (such as RDF and OWL ontologies) [33-35]. * **Industry confusion and risk:** Because of this "semantics of semantics" terminology war, **enterprises risk conflating basic metric governance with actual knowledge representation** [36-38]. * **The need for honesty:** The industry must clearly delineate that BI tools are great for metric consistency, but **true semantic infrastructure needed for performant AI agents requires ontologies and knowledge graphs** [28, 38]. ### "The data reliability question you're avoiding" by Andrew Jones * **Speed over reliability:** Data engineers routinely prioritize moving quickly over building reliable systems. This manifests as **permanent "quick fixes" in ETL processes and shipping pipelines without proper monitoring or alerting** [39]. * **Misaligned expectations:** If users expect highly reliable data to power core business processes or ML-driven products, engineers who deliberately sacrifice reliability for speed are making the wrong trade-offs [39, 40]. * **Relocating rigor:** The introduction of AI-assisted coding does not remove the need for strict engineering standards; **it simply moves where human rigor must be applied** within the data pipeline [40].