B216 words · Upper-intermediate

Artificial Intelligence vocabulary — B2

Artificial intelligence vocabulary such as machine learning, dataset and training data, plus classify, recommendation and guardrail. Words for describing how a system is built.

16 questions

Words in this set

automation
/ɔtəˈmeɪʃən/
The use of machines and programs to do work that people used to do by hand.
Example: Automation has changed how the whole factory runs.
prediction
/priˈdɪkʃən/
A statement about what will probably happen, based on data you already have.
Example: The model's prediction about rainfall was surprisingly close.
recognition
/ˌrɛkəɡˈnɪʃən/
The ability of a system to identify a face, voice or object it has seen before.
Example: Face recognition unlocks my phone in a second.
identify
/aɪˈdɛntəˌfaɪ/
To find out exactly what or who something is, often by comparing it with known examples.
Example: The camera can identify faces in a crowd.
process
/ˈprɑˌsɛs/
To use a computer to handle, change, or study information.
Example: It takes a moment for the app to process your photo.
suggest
/səˈdʒɛst/
To offer an idea or option to someone, often based on things a program already knows about them.
Example: The app will suggest a song you might like.
personalize
/ˈpɜrsənəˌlaɪz/
To change something so that it fits one particular person's needs or tastes.
Example: The app can personalize your news feed based on what you read.
feature
/ˈfitʃər/
One part of a program or device that lets you do a particular thing.
Example: The new feature lets you talk to the app instead of typing.
dataset
/ˈdeɪtəˌsɛt/
A large collection of information that a computer program is trained or tested on.
Example: The model learned from a huge dataset of photos.
machine learning
A way that computer programs improve by studying huge amounts of information, instead of following fixed rules.
Example: Machine learning helps the app get better at picking songs you like.
training data
The examples a system learns from.
Example: The training data decided what the model was good at.
guardrail
/ˈɡɑrˌdreɪl/
A rule built into a system to stop it doing something harmful.
Example: One guardrail blocked the request and three others let it through.
classify
/ˈklæsəˌfaɪ/
To sort things into groups by type.
Example: The model can classify a scan in under a second.
recommendation
/ˌrɛkəmənˈdeɪʃən/
A suggestion a system makes based on what you did before.
Example: The recommendation was uncanny and slightly unwelcome.
query
/ˈkwiri/
A question put to a system in order to get information.
Example: One badly formed query can cost an hour of compute.
compute
/kəmˈpjut/
The processing power needed to run a system.
Example: Compute turned out to cost more than the engineers did.