Hugo Alves
@huguini.bsky.social
Building https://syntheticusers.com
The best thing about this app is watching people talk about what LLMs can and can’t do but their only experience is chatting with ChatGPT on the free version.
December 14, 2024 at 8:09 PM
The best thing about this app is watching people talk about what LLMs can and can’t do but their only experience is chatting with ChatGPT on the free version.
The coolest thing about this app in contrast to the other is that it was able to gather the most annoying people in a single place:
There’s deep thoughts and empathic virtue signalling at every corner.
Glad we have the tech/AI bros on the other one.
There’s deep thoughts and empathic virtue signalling at every corner.
Glad we have the tech/AI bros on the other one.
November 19, 2024 at 9:07 PM
The coolest thing about this app in contrast to the other is that it was able to gather the most annoying people in a single place:
There’s deep thoughts and empathic virtue signalling at every corner.
Glad we have the tech/AI bros on the other one.
There’s deep thoughts and empathic virtue signalling at every corner.
Glad we have the tech/AI bros on the other one.
This morning's #UX read: uxdesign.cc/how-ai-is-fo... had a quote that stood out for me:
'One alternative is Synthetic Users, algorithms capable of analyzing CRM data, social media, purchase history, and other company data, creating behavior groups in real time.'
OK where to start +
'One alternative is Synthetic Users, algorithms capable of analyzing CRM data, social media, purchase history, and other company data, creating behavior groups in real time.'
OK where to start +
How AI is enabling designers to go beyond personas
The creation of Personas has brought us this far, but it seems that it will no longer be enough for the approaching future.
uxdesign.cc
November 19, 2024 at 9:04 PM
Reposted by Hugo Alves
Crazy interesting paper in many ways:
1) Voice-enabled GPT-4o conducted 2 hour
interviews of 1,052 people
2) GPT-4o agents were given the transcripts & prompted to simulate the people
3) The agents were given surveys & tasks. They achieved 85% accuracy in simulating interviewees real answers!
1) Voice-enabled GPT-4o conducted 2 hour
interviews of 1,052 people
2) GPT-4o agents were given the transcripts & prompted to simulate the people
3) The agents were given surveys & tasks. They achieved 85% accuracy in simulating interviewees real answers!
November 19, 2024 at 12:18 AM
Crazy interesting paper in many ways:
1) Voice-enabled GPT-4o conducted 2 hour
interviews of 1,052 people
2) GPT-4o agents were given the transcripts & prompted to simulate the people
3) The agents were given surveys & tasks. They achieved 85% accuracy in simulating interviewees real answers!
1) Voice-enabled GPT-4o conducted 2 hour
interviews of 1,052 people
2) GPT-4o agents were given the transcripts & prompted to simulate the people
3) The agents were given surveys & tasks. They achieved 85% accuracy in simulating interviewees real answers!
Reposted by Hugo Alves
This is wild. User researchers (myself definitely included) scoffed at the idea of doing research with “synthetic users” a year or so ago. Might be time to start thinking differently about the idea.
Crazy interesting paper in many ways:
1) Voice-enabled GPT-4o conducted 2 hour
interviews of 1,052 people
2) GPT-4o agents were given the transcripts & prompted to simulate the people
3) The agents were given surveys & tasks. They achieved 85% accuracy in simulating interviewees real answers!
1) Voice-enabled GPT-4o conducted 2 hour
interviews of 1,052 people
2) GPT-4o agents were given the transcripts & prompted to simulate the people
3) The agents were given surveys & tasks. They achieved 85% accuracy in simulating interviewees real answers!
November 19, 2024 at 3:22 PM
This is wild. User researchers (myself definitely included) scoffed at the idea of doing research with “synthetic users” a year or so ago. Might be time to start thinking differently about the idea.