I’ve seen myself progress through a few “levels” here.
Level 0: tinker with the novelty of the tech but don’t extract real value. This was week 1 and i was mostly just having it create silly stories to make my kids laugh.
Level 1: use LLMs as a replacement for searching the internet. This is where most people get and stay imo.
Level 2: conversations with LLMs that begin to border on collaboration. Eg you ask follow ups.
Level 3: full collaboration. This is when you start to get meaningful value imo. Discuss ideas with the LLM, challenge it, have it challenge you. You can do this anything. On a weekly basis I’m collaborating on plans for date night with my wife, schedules to bake bread with daughter, and anything else going on in my life. I find the answers with the LLM, same way you’d collaborate with a human.
Level 3.5/4: i now regularly run deep research searches. For example I’m going out with friends in a few weeks and want to have a fun night. I’m passively running searches and iterating now despite no time crunch. Basically I’m fully delegating research tasks, giving directions on next steps as needed, and i would say this saves me hours per week.
I’ve essentially stopped using Google in favor of ChatGPT and Perplexity. It’s just excellent for answering simple questions without me having to click or search on some random webpage.
I'm almost always on perplexity now, it has supplanted Google search.
Curious what you're using chat GPT on and not perplexity? I feel that its free tier offers more than ChatGPT. It is particularly helpful for the "tell me how ..." kind of queries and to confirm understanding with concepts I'm learning
Yeah sorry I should’ve been more clear. My absolute preference for Google replacement is Perplexity. Sometimes I’ll use ChatGPT just because I have the app open or a tab open or something and find that it’s OK enough.
I have paid versions of both but think that I get more out of the money for ChatGPT than Perplexity.
> "Enshittification" is a term coined by Cory Doctorow to describe the gradual decline in quality and user experience of a product or service, particularly online platforms, as they prioritize profits over user satisfaction
I used ChatGPT to adjust what exercises I am doing. I asked what exercises impacted what muscles (primary and secondary), put it in a spreadsheet, so I could visualize the overlap. Then asked what muscle groups am I not addressing. Finally what exercises would address those neglected muscles.
I'm moving house and it's great for feeding in floor plans and coming up with ideas. Like having an interior designer. Not always accurate but it's enough to get me going.
(Use Gemini 2.5 pro in Google AI Studio instead. 25 free requests per day using up to 1M tokens each. I honestly keep being amazed about what it's capable of.)
I'm having some issues with Gemini giving me inaccurate information with a lot of confidence. Have you faced this? And if so, how do you ensure you're able to trust its responses.
I’ve seen myself progress through a few “levels” here.
Level 0: tinker with the novelty of the tech but don’t extract real value. This was week 1 and i was mostly just having it create silly stories to make my kids laugh.
Level 1: use LLMs as a replacement for searching the internet. This is where most people get and stay imo.
Level 2: conversations with LLMs that begin to border on collaboration. Eg you ask follow ups.
Level 3: full collaboration. This is when you start to get meaningful value imo. Discuss ideas with the LLM, challenge it, have it challenge you. You can do this anything. On a weekly basis I’m collaborating on plans for date night with my wife, schedules to bake bread with daughter, and anything else going on in my life. I find the answers with the LLM, same way you’d collaborate with a human.
Level 3.5/4: i now regularly run deep research searches. For example I’m going out with friends in a few weeks and want to have a fun night. I’m passively running searches and iterating now despite no time crunch. Basically I’m fully delegating research tasks, giving directions on next steps as needed, and i would say this saves me hours per week.
Asking whatever questions pop into my mind and it gives me a nice overview of it.
Today I was on the bus and was curious about how the many moons of other planets affect those planets..
Basically take “stoner thoughts” and talk to it lol
I’ve essentially stopped using Google in favor of ChatGPT and Perplexity. It’s just excellent for answering simple questions without me having to click or search on some random webpage.
I'm almost always on perplexity now, it has supplanted Google search.
Curious what you're using chat GPT on and not perplexity? I feel that its free tier offers more than ChatGPT. It is particularly helpful for the "tell me how ..." kind of queries and to confirm understanding with concepts I'm learning
Yeah sorry I should’ve been more clear. My absolute preference for Google replacement is Perplexity. Sometimes I’ll use ChatGPT just because I have the app open or a tab open or something and find that it’s OK enough.
I have paid versions of both but think that I get more out of the money for ChatGPT than Perplexity.
How long until AI results/output get the enshittification treatment?
Hopefully atleast 1-2 good years before we start skipping the starting of answers.
It’s coming soon. I can feel it. It will start segueing into NordVPN and BetterHelp ads at the end of its responses.
You know what else could help with this? Would you like me to sign you up for a NordVPN free trial using code OPENAI?
What does “enshittification” mean?
> "Enshittification" is a term coined by Cory Doctorow to describe the gradual decline in quality and user experience of a product or service, particularly online platforms, as they prioritize profits over user satisfaction
- Google Gemini results
Not rocket science, but ...
I used ChatGPT to adjust what exercises I am doing. I asked what exercises impacted what muscles (primary and secondary), put it in a spreadsheet, so I could visualize the overlap. Then asked what muscle groups am I not addressing. Finally what exercises would address those neglected muscles.
End result is my workouts are more productive.
I'm moving house and it's great for feeding in floor plans and coming up with ideas. Like having an interior designer. Not always accurate but it's enough to get me going.
Diving deeper into exercise posture correction and journalling/motivation.
(Use Gemini 2.5 pro in Google AI Studio instead. 25 free requests per day using up to 1M tokens each. I honestly keep being amazed about what it's capable of.)
I'm having some issues with Gemini giving me inaccurate information with a lot of confidence. Have you faced this? And if so, how do you ensure you're able to trust its responses.