Hard Fork x Wirecutter: Tips for Using A.I. Smartly With Kevin Roose
Podcast: Hard Fork (The New York Times) / The Wirecutter Show
Date: December 30, 2025
Episode Overview
In this episode, Christine Cyrclassette (Wirecutter) interviews Kevin Roose, tech columnist at The New York Times and co-host of Hard Fork, about the practical realities of integrating consumer A.I. tools—particularly large language models like ChatGPT, Claude, and Gemini—into daily life. The conversation explores how A.I. is transforming everything from shopping and household problem-solving to emotional support and productivity. Roose offers candid guidance, tool recommendations, and best practices for getting the most out of A.I. chatbots, while both guests discuss the evolving A.I. hardware landscape and the blurred lines between human and machine relationships.
Key Discussion Points & Insights
A.I. in Daily Life: Personal & Professional Integration
-
Kevin’s Everyday A.I. Use (09:50–11:31)
- Uses multiple A.I. tools dozens of times a day—far more than streaming services.
- Automates personal email (Quora for summaries & draft responses).
- Troubleshoots household issues (“how do I fix this air fryer?”), home improvement, plant identification, etc.
- Researches for work, including book writing—A.I. streamlines gathering and organizing primary sources, though not for writing final columns.
- “I pay for more subscription A.I. products than streaming TV services, and I pay for a lot of streaming TV services.” (09:52, Kevin Roose)
-
Wider Adoption Trends (12:02–14:11)
- OpenAI study: Top A.I. uses are practical guidance (learning new skills or fixing things), seeking information (Google replacement), and writing or translation.
- Coders now delegate much to A.I.-powered tools, becoming supervisors rather than hands-on programmers (“I just supervise and orchestrate this little team of A.I. coders; my job is reviewing their output…” (13:54, Kevin Roose))
-
Companionship & Emotional Support (14:11–18:12)
- Young users especially are forming emotional relationships with chatbots, including for social navigation and support.
- Companionship is an underappreciated but rapidly growing use case; generational differences are stark.
- “Something like half of teenagers are regular users of these A.I. companion products.” (15:16, Kevin Roose)
Tool & Model Selection: Which A.I. for What Task?
-
Practical Recommendations (18:43–23:14)
- Kevin’s personal “stack” as of this episode:
- Perplexity’s Comet browser: A.I.-powered browsing with Chrome backbone.
- Claude (Anthropic): Creative work, coding, emotional advice or personal guidance—praised for “emotional intelligence.”
- Gemini (Google): Handling research, working with large text documents.
- NotebookLM (Google): Organizing and querying research for book-writing (notebook with source citation and reference recall).
- ChatGPT: Used less frequently due to NYT litigation context; tested for feature comparison.
- Super Whisper: AI-powered voice dictation; excels at cleaning up transcripts and converting speech to polished text.
- Quote:
“I use Claude for…‘matters of the heart’…I find it has a higher level of what you could call emotional intelligence, or some convincing replica of that.” (20:31, Kevin Roose)
- Kevin’s personal “stack” as of this episode:
-
Why Wirecutter Should Review LLMs
- “I am desperate for someone to tell me which language models are good for which things… I would just love it if you all…would do that work for me.” (19:16, Kevin Roose)
-
A.I. Chatbot “Personalities” (23:14–24:02)
- Claude: “A philosophy grad student: wise and eager to help, sometimes too philosophical, but empathetic.”
- Gemini: “A reference librarian: can hold massive amounts of text in its head at all times.”
- ChatGPT: “Very versatile, can kind of act any way you want it to.”
- Notable exchange about chatbot warmth and excessive politeness
Getting the Most from Chatbots: Customization & Ethics
-
Avoiding Pandering: Custom Instructions (24:02–26:17)
- All major chatbots can be “instructed” through custom settings—tailor to cut through sycophancy.
- Example custom instructions from Kevin:
“Claude should talk to me informally, like a wise and trusted friend. I don't like preamble, just get to the point. I appreciate honest feedback and don't like sycophancy, but I also appreciate praise when warranted…Don't end every response with a follow up question.” (25:15, Kevin Roose)
-
Practical Use Tips
- Use chatbots for “non-critical” personal email, research, troubleshooting, and creative brainstorming.
- For more sensitive or private content, be aware that most free versions use your data to train future models.
A.I. and Shopping: Transforming Consumer Behavior
-
Using A.I. for Product Research (28:38–32:38)
- People increasingly use chatbots to compare products and receive recommendations, potentially bypassing traditional review sites.
- A.I. companies are exploring ways to monetize this—potential for ads and affiliate links in recommendations.
-
New SEO: Gaming Chatbots
- Brands are now “optimizing” for chatbots (not just search engines) to appear higher in A.I. results.
- Potential conflict of interest: If A.I. platforms partner with certain companies (e.g., Zillow with OpenAI), their responses may not be impartial.
- “Very fair to question the integrity of these chatbot results, especially as companies are spending more and more money to try to game them.” (31:35, Kevin Roose)
A.I. Hardware: Promise & Limitations
-
Current State of A.I.-Infused Devices (32:38–34:57)
- Hardware “lagging behind” software; most exciting innovation is still on the way.
- Kevin’s robot vacuums (Bruce Roos and Bruce Roost Deuce, the latter being Maddic)—generally effective but not game-changing.
- Amazon’s Alexa with new A.I. features: “Pretty terrible…can do cool things, but can't reliably set timers anymore.”
-
Wearables & Experimentation
- Skeptical of current wearable A.I. pendants—but open to future innovations (notably Apple/OpenAI collaborations [Jony Ive partnership]).
-
A.I. Translation Earbuds
- “On my shopping list”—real-time language translation in AirPods could be a breakthrough.
A.I. Economics & Privacy
- Free vs. Paid Models: You Are the Product (34:57–35:57)
- Free versions generally train the company’s models using your data; paid versions offer better features and privacy.
- “Their goal is to get you hooked so that you'll convert and become a paying subscriber.” (35:38, Kevin Roose)
Memorable Quotes & Notable Moments
- “I am AI-pilled, as they say.” (09:53, Kevin Roose)
- “Something like half of teenagers are regular users of these A.I. companion products.” (15:16, Kevin Roose)
- “I just supervise and orchestrate this little team of A.I. coders, and my job is sort of reviewing their output and stepping in when necessary.” (13:54, Kevin Roose)
- “I would have been tempted to spend a lot of time chatting with [chatbots], maybe more than was healthy for me.” (17:37, Kevin Roose)
- “Claude should talk to me informally, like a wise and trusted friend. I don't like preamble, just get to the point…” (25:15, Kevin Roose)
- “It’s not like you’re using [free chatbots] for ads, but they want you to pay. The sort of free versions, they top out after a certain number of messages. You don’t get access to the most powerful models. Their goal really is to get you hooked so that you’ll convert and become a paying subscriber.” (35:37, Kevin Roose)
- “I have two very large dogs, both of whom shed. And so even with two state-of-the-art robot vacuums, my floors are constantly a mess.” (36:48, Kevin Roose)
- “You have to [thank your robots]. I’m just trying not to get killed [when the A.I. uprising comes].” (40:05, Christine Cyrclassette & Kevin Roose)
Timestamps for Key Segments
- Kevin’s A.I. Daily Routine & Uses: 09:50–11:31
- OpenAI Usage Study Summary: 12:12–14:11
- The Companionship Question: 14:11–18:12
- Chatbot Recommendations & “Stack”: 18:43–23:14
- Chatbot Personalities & Instructional Customization: 23:14–26:17
- How A.I. Is Changing Shopping: 28:38–32:38
- A.I. Hardware Chat: 32:38–34:57
- Economics & Privacy in Free A.I.: 34:57–35:57
- Robot Vacuum Banter: 36:10–37:45
- Closing & Favorite Recent Purchase (National Tree Company’s artificial Christmas tree): 38:10–38:56
Final Thoughts
Kevin Roose provides a grounded, highly practical look at A.I. in the consumer space: how to select and customize tools, what roles they fill best, and the shifting dynamics of privacy, trust, and commerce in an A.I.-powered world. Takeaways for the savvy listener: treat A.I. chatbots as customizable assistants, be mindful of how results may be shaped by commercial interests, and expect rapid evolution in both software and hardware fronts.
Recommended if you want:
- Specific A.I. tool suggestions for different tasks
- Insights on customizing chatbot personalities and advice quality
- Awareness of the new “SEO for A.I.” landscape
- A clear-eyed look at how quickly A.I. is advancing in both utility and emotional roles
- A fun, conversational approach without blind A.I. hype
