Archives des Computer Science and Artificial Intelligence Laboratory (CSAIL)

Can AI really code? Study maps the roadblocks to autonomous software engineering
19 Juil
AI shapes autonomous underwater “gliders”
19 Juil
Study could lead to LLMs that are better at complex reasoning
19 Juil
Using generative AI to help robots jump higher and land safely
19 Juil
Researchers present bold ideas for AI at MIT Generative AI Impact Consortium kickoff event
19 Juil
Unpacking the bias of large language models
19 Juil
Envisioning a future where health care tech leaves some behind
19 Juil
Teaching AI models what they don’t know
19 Juil
Teaching AI models the broad strokes to sketch more like humans do
19 Juil
AI learns how vision and sound are connected, without human intervention
19 Juil
Study shows vision-language models can’t handle queries with negation words
19 Juil
Hybrid AI model crafts smooth, high-quality videos in seconds
19 Juil
Novel AI model inspired by neural dynamics from the brain
19 Juil
Making AI models more trustworthy for high-stakes settings
19 Juil
The MIT-Portugal Program enters Phase 4
19 Juil
“Periodic table of machine learning” could fuel AI discovery
19 Juil
3D modeling you can feel
19 Juil
Making AI-generated code more accurate in any language
19 Juil
Training LLMs to self-detoxify their language
19 Juil
New method efficiently safeguards sensitive AI training data
19 Juil
Could LLMs help design our next medicines and materials?
19 Juil
New method assesses and improves the reliability of radiologists’ diagnostic reports
19 Juil
For this computer scientist, MIT Open Learning was the start of a life-changing journey
19 Juil
At the core of problem-solving
19 Juil
Robotic helper making mistakes? Just nudge it in the right direction
19 Juil
Like human brains, large language models reason about diverse data in a general way
19 Juil
AI model deciphers the code in proteins that tells them where to go
19 Juil
Creating a common language
19 Juil
Validation technique could help scientists make more accurate forecasts
19 Juil
Introducing the MIT Generative AI Impact Consortium
19 Juil
User-friendly system can help developers build more efficient simulations and AI models
19 Juil
3 Questions: Modeling adversarial intelligence to exploit AI’s security vulnerabilities
19 Juil
Toward video generative models of the molecular world
19 Juil
Explained: Generative AI’s environmental impact
19 Juil
Teaching AI to communicate sounds like humans do
19 Juil
A new computational model can predict antibody structures more accurately
19 Juil
Ecologists find computer vision models’ blind spots in retrieving wildlife images
19 Juil
MIT researchers introduce Boltz-1, a fully open-source model for predicting biomolecular structures
19 Juil
Study reveals AI chatbots can detect race, but racial bias reduces response empathy
19 Juil
Lara Ozkan named 2025 Marshall Scholar
19 Juil
MIT affiliates named 2024 Schmidt Futures AI2050 Fellows
19 Juil
Teaching a robot its limits, to complete open-ended tasks safely
19 Juil
AI in health should be regulated, but don’t forget about the algorithms, researchers say
19 Juil

Tag

#Computer Science and Artificial Intelligence Laboratory (CSAIL)

Can AI really code? Study maps the roadblocks to autonomous software engineering

AI shapes autonomous underwater “gliders”

Study could lead to LLMs that are better at complex reasoning

Using generative AI to help robots jump higher and land safely

Unpacking the bias of large language models

Teaching AI models what they don’t know

Teaching AI models the broad strokes to sketch more like humans do

AI learns how vision and sound are connected, without human intervention

Study shows vision-language models can’t handle queries with negation words

Hybrid AI model crafts smooth, high-quality videos in seconds

Novel AI model inspired by neural dynamics from the brain

Making AI models more trustworthy for high-stakes settings

The MIT-Portugal Program enters Phase 4

“Periodic table of machine learning” could fuel AI discovery

3D modeling you can feel

Making AI-generated code more accurate in any language

Training LLMs to self-detoxify their language

New method efficiently safeguards sensitive AI training data

Could LLMs help design our next medicines and materials?

New method assesses and improves the reliability of radiologists’ diagnostic reports

For this computer scientist, MIT Open Learning was the start of a life-changing journey

At the core of problem-solving

Robotic helper making mistakes? Just nudge it in the right direction

Like human brains, large language models reason about diverse data in a general way

AI model deciphers the code in proteins that tells them where to go

Creating a common language

Validation technique could help scientists make more accurate forecasts

User-friendly system can help developers build more efficient simulations and AI models

3 Questions: Modeling adversarial intelligence to exploit AI’s security vulnerabilities

Toward video generative models of the molecular world

Explained: Generative AI’s environmental impact

Teaching AI to communicate sounds like humans do

A new computational model can predict antibody structures more accurately

Ecologists find computer vision models’ blind spots in retrieving wildlife images

MIT researchers introduce Boltz-1, a fully open-source model for predicting biomolecular structures

Lara Ozkan named 2025 Marshall Scholar

Teaching a robot its limits, to complete open-ended tasks safely

AI in health should be regulated, but don’t forget about the algorithms, researchers say

Can AI really code? Study maps the roadblocks to autonomous software engineering

AI shapes autonomous underwater “gliders”

Study could lead to LLMs that are better at complex reasoning

Using generative AI to help robots jump higher and land safely

Researchers present bold ideas for AI at MIT Generative AI Impact Consortium kickoff event

Unpacking the bias of large language models

Envisioning a future where health care tech leaves some behind

Teaching AI models what they don’t know

Teaching AI models the broad strokes to sketch more like humans do

AI learns how vision and sound are connected, without human intervention

Study shows vision-language models can’t handle queries with negation words

Hybrid AI model crafts smooth, high-quality videos in seconds

Novel AI model inspired by neural dynamics from the brain

Making AI models more trustworthy for high-stakes settings

The MIT-Portugal Program enters Phase 4

“Periodic table of machine learning” could fuel AI discovery

3D modeling you can feel

Making AI-generated code more accurate in any language

Training LLMs to self-detoxify their language

New method efficiently safeguards sensitive AI training data

Could LLMs help design our next medicines and materials?

New method assesses and improves the reliability of radiologists’ diagnostic reports

For this computer scientist, MIT Open Learning was the start of a life-changing journey

At the core of problem-solving

Robotic helper making mistakes? Just nudge it in the right direction

Like human brains, large language models reason about diverse data in a general way

AI model deciphers the code in proteins that tells them where to go

Creating a common language

Validation technique could help scientists make more accurate forecasts

Introducing the MIT Generative AI Impact Consortium

User-friendly system can help developers build more efficient simulations and AI models

3 Questions: Modeling adversarial intelligence to exploit AI’s security vulnerabilities

Toward video generative models of the molecular world

Explained: Generative AI’s environmental impact

Teaching AI to communicate sounds like humans do

A new computational model can predict antibody structures more accurately

Ecologists find computer vision models’ blind spots in retrieving wildlife images

MIT researchers introduce Boltz-1, a fully open-source model for predicting biomolecular structures

Study reveals AI chatbots can detect race, but racial bias reduces response empathy

Lara Ozkan named 2025 Marshall Scholar

MIT affiliates named 2024 Schmidt Futures AI2050 Fellows

Teaching a robot its limits, to complete open-ended tasks safely

AI in health should be regulated, but don’t forget about the algorithms, researchers say