Why AI is making you less productive (and what to do about it)
Here is this week's digest:
Ask HN: You have $35,000 how do you invest it?
When faced with investing $35,000 outside of typical stocks, bonds, crypto, or real estate, a discussion revealed several creative strategies. Key ideas include starting tangible small businesses like a coffee stand, laundromat, or a niche service like car flipping with an emphasis on honest practices. Another popular avenue is investing in specialized physical assets such as machine tools for a small factory, or collectibles like vintage items, classic cars, or even freeze-dried food, leveraging specific knowledge for arbitrage. Some also advocated for self-investment, using the funds for skill development, personal growth, or dedicating time to build one's own venture. The core theme is active participation and finding value in less conventional, often tangible, assets or services.
Ask HN: How can I raise $8k to build a classroom in rural Tajikistan?
An individual seeking to raise $8,000 for a classroom in rural Tajikistan received advice on fundraising, storytelling, and gaining visibility. Key suggestions included using platforms like GoFundMe, leveraging social media, and considering the building's multi-purpose potential (e.g., as a community center). The importance of community involvement was highlighted, with parents planning to cover teacher salaries. However, the discussion also underscored a significant challenge: skepticism arose due to commenters suspecting the use of AI-generated images on the project's GoFundMe page, emphasizing the critical need for transparency and authentic representation in online fundraising efforts.
Ask HN: Is AI 'context switching' exhausting?
Developers discuss the mental exhaustion that can arise from 'context switching' between manually coding and using AI tools. Key reasons cited include disrupted flow states, the cognitive load of correcting AI, dopamine crashes from chasing ideal results, and the simple act of waiting for AI responses.
Helpful strategies shared include:
- Treating AI like an asynchronous junior developer: delegate tasks and work on something else.
- Batching AI work: using it for small, specific tasks like generating snippets or boilerplate.
- Setting clear boundaries: AI for grunt work, not core logic.
- Breaking down large AI tasks into smaller, concurrent ones.
- Using AI in an 'ask mode' where it suggests rather than directly implements code.
Ask HN: How to regain the ability to read with focus and learn
To regain the ability to focus on reading, many suggest starting with something enjoyable and easy, like a fiction series, to rebuild the reading 'muscle'. The key is consistent practice, even if it's just for 15-30 minutes a day. Creating a distraction-free environment is crucial—put your phone in another room or use a dedicated e-reader with Wi-Fi off.
Other effective strategies include:
- Active Reading: Engage with the text by taking handwritten notes, summarizing paragraphs in your own words, and analyzing the author's structure. The book How to Read a Book by Mortimer Adler is a recommended resource for this.
- Lifestyle Changes: Poor focus can be a symptom of other issues. Consider getting an eye test, improving sleep hygiene, meditating, reducing alcohol or caffeine, and managing stress.
- Social Motivation: Joining a book club or seminar can provide the accountability needed to read thoroughly and be prepared for discussion.
Ask HN: Using AI daily but not seeing productivity gains – is it just me?
Many developers are using AI tools daily but aren't seeing the promised productivity leaps. The consensus from a recent discussion suggests the most significant gains come from targeted, specific use cases rather than open-ended prompting.
Key strategies for success include:
- Prefer Autocomplete: Tools like GitHub Copilot are often praised as the 'sweet spot'. They save time on boilerplate and repetitive code without the friction of 'agentic' workflows, which can turn the developer into a full-time code reviewer correcting mistakes.
- Write Detailed Specs: Instead of general prompts, provide LLMs with detailed specifications. This 'spec-first' approach directs the AI more effectively, though some argue that iterating on code (the ultimate spec) with context is more efficient.
- Focus on Isolated Tasks: AI excels at generating self-contained code, like creating an API wrapper from documentation or writing a utility function. It's less effective when deep context of a large, existing codebase is required.
- Target High-Friction, Boring Tasks: Use AI to magnify your workflow, not replace it. Focus it on tedious but necessary tasks like refactoring small code blocks, drafting documentation, or generating test data. The key is to direct the AI on tasks where the time to direct is less than the time to do.
Ask HN: How to get rid of Gemini?
Users frustrated with Google's forced integration of its Gemini AI are sharing ways to disable it and find alternatives. For Google Search, the most effective trick is adding the udm=14 parameter to the search URL, which forces a return to the classic "Web" results view. Websites like udm14.com and browser extensions can automate this process. Alternatively, ad-blockers like uBlock Origin can be used to hide the AI overview panel by blocking its CSS selector (e.g., .hdzaWe).
For those looking to switch search engines, DuckDuckGo was a popular suggestion, with its founder noting the existence of an AI-free version at noai.duckduckgo.com. Kagi was also highly recommended as a paid, ad-free alternative. The discussion highlights that while disabling AI in search is manageable, removing it from integrated products like Google Workspace is far more difficult, pushing many to consider a complete, albeit challenging, migration away from the Google ecosystem.
Add a comment: