- Practice
- About 15 min
- 5 practice questions
Matching the technology to the use: AI, machine learning, big data and cloud
Read a short case and name the technology at work - AI, a type of machine learning, a big-data 'V' or a cloud service model - and avoid the usual mix-ups.
If something looks wrong, please tell us.
Exam questions on future skills rarely ask for a definition alone. They describe a situation - an app, an office, a farm - and ask which technology it shows. This lesson trains that skill for the first group of technologies: AI, machine learning, big data and cloud computing.
Clue words to technology
| Clue in the case | Technology | Watch out for |
|---|---|---|
| Understands speech, recognises faces, translates, answers like a person | Artificial intelligence | A fixed calculator or timer is not AI |
| Learns from past data, gets better with more examples | Machine learning | ML is a part of AI, not separate from it |
| Learns from examples with correct answers attached | Supervised learning | Labels are the clue |
| Finds groups with no answers given | Unsupervised learning | 'Clusters', 'segments' are the clue |
| Learns by trial, reward and penalty | Reinforcement learning | Games and robots are common examples |
| Huge amount / very fast / many forms of data | Big data: volume / velocity / variety | Match the exact V to the clue |
| Rent servers or storage over the internet | IaaS | You still manage the software |
| Ready platform to build your own app | PaaS | You manage only your app |
| Finished software used in a browser | SaaS | Nothing to install or manage |
A quick method for 'which V?' questions
- If the case stresses how much data - Volume.
- If it stresses how fast data arrives or must be processed - Velocity.
- If it stresses different kinds of data (text, photos, video, sensor readings) - Variety.
- If it stresses whether the data can be trusted - Veracity.
- If it stresses the benefit gained - Value.
Worked example 1
Case: An e-commerce company groups its customers into segments by their buying habits, without telling the computer what the groups should be. Working: learning from data = ML; no labels given, the computer forms groups = unsupervised learning. Trap: supervised learning would need labelled examples.
Worked example 2
Case: A state transport department analyses GPS signals, ticket sales, CCTV images and passenger complaints together to plan bus routes. Working: the stress is on many different forms of data - GPS numbers, images, text. Answer: variety. If the case had said 'millions of records' it would point to volume; 'updated every second' would point to velocity.
Worked example 3
Case: A school uses an online office suite for documents and spreadsheets through a browser, with nothing installed and no servers to manage. Working: complete software delivered over the internet = SaaS. If the school's IT teacher instead rented a virtual server and installed software on it, that would be IaaS.
Common mix-ups
- 'Every app is AI' - no; only those doing tasks needing human-like intelligence.
- 'Cloud means no internet needed' - the opposite; cloud is reached over the internet.
- 'Public cloud means anyone can see your files' - no; public means shared infrastructure offered to the public, with each customer's data kept separate and protected.
- 'Machine learning programs are fully written as rules by hand' - no; they learn the rules from data.
Ask: does it think (AI)? does it learn from data (ML)? which V is stressed? who manages what (IaaS, PaaS, SaaS)?
Key takeaways: turn each case into a clue word, then into a technology. Labels mean supervised learning, no labels unsupervised, rewards reinforcement. Match the stressed quality of data to one V. In the cloud, the less you manage, the further you are along IaaS, PaaS, SaaS.
Check your understanding
-
1. A program learns to play chess better by getting a reward for wins and a penalty for losses. This is:
- Supervised learning
- Reinforcement learning
- Unsupervised learning
- Robotic process automation
Show answer
Answer: B. Learning by trial and error from rewards and penalties is reinforcement learning.
-
2. A case stresses that sensor readings arrive every second and must be analysed immediately. Which V of big data is stressed?
- Velocity
- Variety
- Volume
- Veracity
Show answer
Answer: A. Velocity is the speed at which data arrives and must be processed.
-
3. A company rents virtual servers over the internet and installs and manages its own operating system and software on them. This is:
- SaaS
- PaaS
- IaaS
- IoT
Show answer
Answer: C. Infrastructure as a Service gives raw computing resources; the customer manages the software on top.
-
4. Which is NOT an example of artificial intelligence?
- A system recognising faces in photos
- An app translating speech between languages
- A chatbot answering customer questions
- A wall clock showing the time
Show answer
Answer: D. A clock follows a fixed mechanism and needs no human-like intelligence. The others recognise, translate or converse.
-
5. Which statement is correct?
- Machine learning is a branch of artificial intelligence
- Cloud computing works without internet
- SaaS means renting bare servers
- Unsupervised learning needs labelled answers
Show answer
Answer: A. ML is part of AI. Cloud needs the internet, SaaS is finished software, and unsupervised learning works without labels.
Sign in to keep track of the lessons you finish.