Adriano Belisario
@belisario.website
It was great to present this work last week at #CSCW2025 in Norway and get valuable feedback from colleagues.
If you're interested in AI applications for humanitarian contexts, I invite you to read the paper and reach out for a chat!
If you're interested in AI applications for humanitarian contexts, I invite you to read the paper and reach out for a chat!
October 27, 2025 at 9:17 PM
It was great to present this work last week at #CSCW2025 in Norway and get valuable feedback from colleagues.
If you're interested in AI applications for humanitarian contexts, I invite you to read the paper and reach out for a chat!
If you're interested in AI applications for humanitarian contexts, I invite you to read the paper and reach out for a chat!
🔹Evaluation should go beyond lab metrics🔹
AI models can ace benchmarks and still flop in real-world settings. How people use AI tools (and the impact these actually have) can be very different from what we expect.
To address this, we assessed the performance through a controlled intervention.
AI models can ace benchmarks and still flop in real-world settings. How people use AI tools (and the impact these actually have) can be very different from what we expect.
To address this, we assessed the performance through a controlled intervention.
October 27, 2025 at 9:17 PM
🔹Evaluation should go beyond lab metrics🔹
AI models can ace benchmarks and still flop in real-world settings. How people use AI tools (and the impact these actually have) can be very different from what we expect.
To address this, we assessed the performance through a controlled intervention.
AI models can ace benchmarks and still flop in real-world settings. How people use AI tools (and the impact these actually have) can be very different from what we expect.
To address this, we assessed the performance through a controlled intervention.
🔹Small, open-source models matter🔹
Proprietary LLMs gets all the hype, but models like BERT variants are still the workhorses of many applications.
Equating “AI” with “LLMs” is not only misleading but also distracts us from alternatives that could greatly benefit low-resourced organizations.
Proprietary LLMs gets all the hype, but models like BERT variants are still the workhorses of many applications.
Equating “AI” with “LLMs” is not only misleading but also distracts us from alternatives that could greatly benefit low-resourced organizations.
October 27, 2025 at 9:17 PM
🔹Small, open-source models matter🔹
Proprietary LLMs gets all the hype, but models like BERT variants are still the workhorses of many applications.
Equating “AI” with “LLMs” is not only misleading but also distracts us from alternatives that could greatly benefit low-resourced organizations.
Proprietary LLMs gets all the hype, but models like BERT variants are still the workhorses of many applications.
Equating “AI” with “LLMs” is not only misleading but also distracts us from alternatives that could greatly benefit low-resourced organizations.
We used a BERT-based model and the Twitter API (RIP 🪦) to classify tweets in real time and, although much has changed since then, the paper still offers important insights. 👇
October 27, 2025 at 9:17 PM
We used a BERT-based model and the Twitter API (RIP 🪦) to classify tweets in real time and, although much has changed since then, the paper still offers important insights. 👇
Fantastic, congrats Scott!
October 27, 2025 at 7:31 PM
Fantastic, congrats Scott!
Já usei sim lá e a do Kaggle, que pelo que lembro tem um limite gratuito bom. Ambos rolaram bem
September 1, 2025 at 3:09 AM
Já usei sim lá e a do Kaggle, que pelo que lembro tem um limite gratuito bom. Ambos rolaram bem
July 19, 2025 at 1:01 PM
Comentando aqui pra entrar na fila também e não perder o anúncio da data!
June 16, 2025 at 1:44 AM
Comentando aqui pra entrar na fila também e não perder o anúncio da data!
Também já tive experiências muito ruins com Rappi e isso de não poder acionar ninguém me fez odiar mt esse app e jurar jamais usar novamente
February 9, 2025 at 10:18 PM
Também já tive experiências muito ruins com Rappi e isso de não poder acionar ninguém me fez odiar mt esse app e jurar jamais usar novamente
Baita casal!*
* Vale para os dois casais
* Vale para os dois casais
October 4, 2024 at 9:56 PM
Baita casal!*
* Vale para os dois casais
* Vale para os dois casais
Como falaram aqui, é baseado na ideia de percepção subliminar. Tem estudos sobre o uso dessas técnicas na comunicação audiovisual, mas não sei se há consenso sobre sua eficácia.
en.wikipedia.org/wiki/Psychor...
en.wikipedia.org/wiki/Psychor...
Psychorama - Wikipedia
en.wikipedia.org
September 13, 2024 at 2:49 PM
Como falaram aqui, é baseado na ideia de percepção subliminar. Tem estudos sobre o uso dessas técnicas na comunicação audiovisual, mas não sei se há consenso sobre sua eficácia.
en.wikipedia.org/wiki/Psychor...
en.wikipedia.org/wiki/Psychor...