Your phone knows your face
You look at your phone and it unlocks. You say "play music" and it plays. A photo app groups all your pictures of a friend.
Ten years ago, this was rare. Now it is normal. These are new technologies, and they change what software developers build, and how.
Key words
| Word | What it means | Example |
|---|---|---|
| Artificial intelligence | Computer systems that do tasks which usually need human thinking. | recognising speech |
| Machine learning | A kind of AI that learns from examples instead of fixed rules. | a spam filter (a tool that blocks junk messages) that learns from past emails |
| Generative AI | AI that makes new text, pictures or code. | a chat tool that writes an email |
| Conversational AI | AI you can talk or type to, like a chat. | a help chat on a shop website |
| Computer vision | AI that understands pictures and video. | reading a number plate |
| Object recognition | Finding and naming things in a picture. | finding a dog in a photo |
| Biometrics | Using your body to prove who you are. | fingerprint, face, voice |
| Data lake | A big store of raw data (data not yet organised) in many shapes. | photos, logs, tables, all kept as they came |
| Data warehouse | A store of tidy, organised data for reports. | monthly sales tables |
The cells students fill in are left blank in this preview.
The technologies at a glance
| Technology | What it does | Example | What it means for developers |
|---|---|---|---|
| Artificial intelligence | Does tasks that need human thinking | A game opponent | New features, and new risks to test |
| Virtual intelligence | Acts like a helper but follows set rules and does not learn | A shop assistant that answers set questions | Developers write the rules and the answers |
| Conversational AI | Talks with people | A bank chat helper | Needs clear replies, and a plan for when it does not understand |
| Generative AI | Makes new text, pictures or code | A code helper | Developers must check what it writes |
| Machine learning | Learns from examples | Film suggestions | Needs good data, and fair data |
| Computer vision and object recognition | Understands pictures | Reading car number plates | Needs cameras and training pictures |
| Biometrics | Proves who you are by your body | Fingerprint login | Must protect very private data |
The cells students fill in are left blank in this preview.
Quick check
Have a go
Quick check
3 questionsYour answers are marked here and are not kept. A college account keeps every student’s marks.
Try it: a program that learns
spam = ["win free prize", "free money now", "claim your prize"]
normal = ["lunch at one", "homework is due", "free time at lunch"]
spam_words = set()
for message in spam:
spam_words.update(message.split())
normal_words = set()
for message in normal:
normal_words.update(message.split())
# A set keeps each word once. The minus sign takes away the normal words.
learned = spam_words - normal_words
print(learned)
test = "you win a prize"
found = False
for word in test.split():
if word in learned:
found = True
print(found)
# Add a new spam message to the list and run it again.
The more examples, the better
data-callouts.note.label
The program did not get rules from a person. It found the words in the examples. With more examples it learns more. With poor or unfair examples, it learns poor or unfair things.
Where does all the data live?
AI needs lots of data. Companies keep it in two main ways.
A data lake keeps data in its raw shape: photos, messages, logs and tables. A data warehouse keeps tidy, organised data, ready for reports.
Lakes are flexible. Warehouses are fast and clear.