How to Keep Up with AI Changes Without Draining Your Energy

The conversation begins with a limit that no new model removes: a person has a finite amount of time and attention. Evgeny starts by deciding what matters to him, then uses those priorities to filter incoming information. Career goals act as a compass. Some updates support that direction; others can be ignored. He plans his days and weeks, while his broader planning horizon has shortened to one or two months. Reading or listening is only part of learning. Discussing a paper with an agent helps him question the argument and notice points he missed. This extends familiar task-management habits rather than replacing them with an entirely new productivity system. Agents can take over recurring preparation, while the person still chooses the subject and decides how much attention it deserves.

Evgeny describes gathering material on selected topics, turning it into audio for walks, and saving useful passages from podcasts. But the knowledge system can itself become a distraction. After spending considerable effort configuring Obsidian, he returned to ordinary Markdown files that an agent helps update and connect. Alexander adds feedback to the selection process. He wants information that expands his understanding rather than merely confirming existing views. His informal technology radar separates topics to read now, topics to monitor, and topics to stop following. Material from different projects can become a research question or an article; a new question can also become something to track. Work, family, and other commitments define the available learning time. The objective is not exhaustive coverage. It is to make deliberate exclusions and revise priorities when new questions emerge. A long morning digest matters less than what it helped someone understand or apply.

Aleksey follows a different route. He switched off automated news delivery because he was no longer reading it. Topics now reach him through a few channels, professional contacts, and the companies he works with. An audit, a training session, or a tool rollout creates a specific question worth investigating. He tries a new model on his own outstanding tasks and compares his experience with other people’s reports. In his experience, too much external commentary can sometimes crowd out an idea before he has formulated it himself. Walks, a change of activity, and writing give an unclear intuition time to develop. A tentative hypothesis does not need to become a publication immediately: it may become clearer after a pause or a conversation. This is a personal practice, not evidence that news or courses hinder everyone. Its purpose is to preserve room for original questions alongside the consumption of ready-made explanations.

Alexander illustrates another limit through his personal learning program. An agent selects theory, videos, and practical exercises and reminds him to study, working upward from data-center infrastructure to models and platforms. Yet the prepared workload can exceed his energy. When a long document becomes difficult to follow or he runs out of attention in the evening, he asks why the subject belongs in the plan and moves into discussion or practice. Recurring operational checks can be automated once data access, change rules, and verification are established. Learning something new still requires participation. The three approaches converge here: handwritten diagrams, notes in one’s own words, discussion, and explaining a topic to someone else help process information. Evgeny uses preparing a talk as a reason to understand a subject properly, and running without a podcast as time to clear his head. Watching many lectures does not provide the experience of building a system and testing how it fails. An agent can prepare inputs or render an idea, but it cannot do the learner’s understanding for them.

The second half turns to the fear of missing opportunities. For Alexander, a major shift was the move from AI that assists with code to agents that can complete substantial tasks independently. He places his own reassessment in late 2025, when those capabilities prompted him to reconsider management, engineering, and his role. The pace of his personal experiments gradually diverged from what he could introduce at work. Choosing to focus on the area that interested him reduced that tension, although the fear of falling behind did not disappear. What helps him is a concrete route: select a domain, draw on accumulated experience, and work toward enough expertise to participate in it. This is an account of a personal decision, not advice that listeners should leave their jobs. The point is to align activity with what the person values. Even abundant free time does not make the study workload infinitely expandable. Interest does not turn human attention into an unlimited resource.

Evgeny suggests comparing today’s self with yesterday’s rather than with the strongest person in a social feed. His evening diary distinguishes a full day from a day that advanced a goal. If meetings and coordination repeatedly produce only busyness, that is a reason to reassess the way he works. He also acknowledges an unresolved limitation: agents make it possible to run more processes in parallel than he has energy to supervise. Experience with context switching helps, but does not remove the need for a life outside work. Aleksey supports refusing to simulate progress; resting can be more useful than continuing activity with no connection to an outcome. The closing argument combines focus with a willingness to change. A main direction and fallback options can be revised after a significant development while the rules for allocating attention remain useful. Adding “yet” to “I cannot do this” keeps learning possible, but a topic deserves effort when it fits the chosen direction. Sustainable progress means continuing purposeful work, not matching every announcement.