Overview
Domain 1 is the vocabulary section of the 1Z0-1122-26 exam. It asks what artificial intelligence means, why organisations use it, and which kinds of data different AI tasks rely on: structured tables, free text, images, audio and time series. You also need the nesting relationship that runs through the rest of the course: AI is the broad goal of machines performing tasks that need human-like intelligence, machine learning is the subset that learns patterns from data, and deep learning is the subset of ML built on layered neural networks. Generative AI sits inside deep learning.
Status note: this page targets the 2026 release, 1Z0-1122-26, which secondary sources describe as the replacement for 1Z0-1122-25 from mid-2026. We could not read Oracle's exam page to confirm the switchover date, so check the code shown in Oracle MyLearn before you book.
How it's tested
Expect one-line definitions and short business examples. A question may describe a use case and ask which AI task or data type it involves, or present a statement about AI, ML and DL and ask which one is correct. Wording is plain, so the risk is overthinking rather than lack of knowledge.
-
Name the AI task
A scenario such as transcribing calls or spotting defects on a production line is given, and you choose the matching task: speech, vision or language.
-
Classify the data
You decide whether the input is structured, unstructured text, image, audio or time-series data and what that implies for the model.
-
Place it in the hierarchy
A statement about AI, machine learning, deep learning or generative AI is tested for which layer it actually describes.
A retailer wants software that reads product reviews and tags each one as positive or negative. The input is unstructured text and the output is a label, so this is a language task built on machine learning. It is not generative AI, because nothing new is produced, and it is not computer vision, because no images are involved. Picking language AI with a classification output is the expected reasoning.
Objectives 3
Official objectives from the Oracle outline, in outline order.
1.1 Explain AI basicsWhat AI is and what it is expected to do.
- What it means
- Artificial intelligence is the ability of a machine to imitate human intellectual work: learning, reasoning, recognising patterns, understanding language and making decisions. Oracle's course frames it around tasks a person would normally need intelligence to perform.
- What to understand
- Know the difference between narrow AI, which handles a specific task such as translation, and artificial general intelligence, a hypothetical system that could handle any intellectual task. Every system on this exam, including large language models, counts as narrow AI. Also know why AI is adopted: automating repetitive work, handling volumes people cannot, and finding patterns hidden in large datasets.
- How it shows up
- Expect a definition question or one asking which statement correctly describes general versus narrow AI.
- Common confusion
- Believing a capable chatbot is general intelligence; it is still a narrow system trained for language tasks.
1.2 Discuss AI applications and types of dataLanguage, audio and vision tasks and the data they consume.
- What it means
- AI applications are grouped by the data they process. Language AI works on text, speech and audio AI works on sound, and vision AI works on images and video. Other tasks, such as forecasting or anomaly detection, work on numeric time-series and tabular data.
- What to understand
- Structured data fits rows and columns, while text, images and audio are unstructured and need conversion into numbers before a model can use them. Text is split into tokens, images become pixel arrays, and audio becomes sampled waveforms or spectrograms. Typical applications include chatbots, translation, transcription, image classification, fraud detection and product recommendation, and you should be able to match each to its data type.
- How it shows up
- A use case is described and you choose the data type or AI task category it belongs to.
- Common confusion
- Treating scanned documents as text data; until OCR runs they are images, which is why document AI uses vision.
1.3 Explain AI vs ML vs DLHow the three terms nest inside each other.
- What it means
- AI is the broad field. Machine learning is the part of AI where systems learn from data instead of following hand-written rules. Deep learning is the part of machine learning that uses neural networks with many layers to learn features directly from raw data.
- What to understand
- Classic ML usually depends on features that people engineer, such as the length of an email or the number of links in it. Deep learning learns those features itself, which is why it suits images, audio and text but needs more data and compute. A rule-based expert system is AI without ML. Generative AI is a branch of deep learning that produces new content rather than only predicting labels or numbers.
- How it shows up
- Statements such as 'all machine learning is deep learning' appear, and you identify which nesting claim holds.
- Common confusion
- Reversing the hierarchy and calling AI a subset of machine learning.
High-yield concepts
The terms, rules and formulas this domain turns into questions most often.
-
Artificial intelligence
Machines performing tasks that need human-like intelligence.
- Remember
- AI is the umbrella term containing ML and DL.
- Tested as
- Definition and hierarchy questions.
- Trap
- Assuming AI always means learning from data; rule-based systems are AI too.
-
Narrow AI
AI built for one task or a small family of tasks.
- Remember
- All current production AI, LLMs included, is narrow.
- Tested as
- Contrasted with general intelligence.
- Trap
- Calling a multi-skill chatbot general AI.
-
Artificial general intelligence
A hypothetical system able to perform any intellectual task a person can.
- Remember
- It does not exist today.
- Tested as
- Used as a distractor in definition items.
- Trap
- Treating AGI as a product category you can deploy.
-
Machine learning
Systems that improve at a task by learning patterns from data.
- Remember
- Learns from examples instead of explicit rules.
- Tested as
- Core to the hierarchy questions.
- Trap
- Mixing it up with traditional programming driven by if-then rules.
-
Deep learning
ML using multi-layer neural networks that learn features automatically.
- Remember
- Needs large datasets and usually GPUs.
- Tested as
- Linked to images, speech and text tasks.
- Trap
- Thinking deep learning removes the need for training data.
-
Structured data
Data organised in rows and columns with a fixed schema.
- Remember
- Classic ML handles it well.
- Tested as
- Data-type matching items.
- Trap
- Calling JSON logs or emails structured just because they are stored in a database.
-
Unstructured data
Text, images, audio and video without a fixed tabular layout.
- Remember
- Must be converted to numeric form for models.
- Tested as
- Data-type and use-case questions.
- Trap
- Assuming unstructured data cannot be used by ML at all.
-
Language AI
AI that understands or produces human language in text form.
- Remember
- Tasks include translation, sentiment and summarisation.
- Tested as
- Maps to OCI Language and LLMs later.
- Trap
- Placing speech-to-text under language rather than speech AI.
-
Vision AI
AI that interprets images and video.
- Remember
- Tasks include classification, detection and OCR.
- Tested as
- Maps to OCI Vision and Document Understanding.
- Trap
- Forgetting that reading printed text from an image starts as a vision task.
-
Speech and audio AI
AI that processes sound, such as transcription or voice synthesis.
- Remember
- Input is a waveform, not text.
- Tested as
- Maps to OCI Speech.
- Trap
- Assuming audio files are just long text files.
Side by side
Concepts candidates mix up, compared on the point that decides the answer.
AI, ML, DL and generative AI
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| AI | The overall goal of machine intelligence. | Includes rule-based systems that never learn. | Treating it as a synonym for ML. |
| Machine learning | Learning patterns from data. | Often relies on engineered features. | Assuming every ML model is a neural network. |
| Deep learning | Layered neural networks learning features from raw data. | Better with unstructured data at scale. | Thinking it works well on tiny datasets. |
| Generative AI | Deep learning models that produce new content. | Outputs text, images or audio rather than a label. | Calling any prediction model generative. |
Data types
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Tabular | Rows and columns with typed fields. | Ready for classic ML algorithms. | Overlooking it as the most common business data. |
| Text | Sequences of words and characters. | Tokenised before modelling. | Ignoring that order and context carry meaning. |
| Image | Grids of pixel values. | Spatial patterns matter. | Treating a scanned form as already searchable text. |
| Audio | Sound sampled over time. | Time-based like text, but continuous. | Assuming it needs no preprocessing. |
Exam traps
Each trap is an answer that looks right. Learn the clue that should change your mind.
-
General versus narrow
- What candidates pick
- Artificial general intelligence
- Why it looks right
- Modern chatbots seem to do everything.
- Why it's wrong
- Every deployed system, including the largest LLMs, is trained for bounded tasks and counts as narrow AI.
Clue to watch for The word 'currently deployed' or a named product.
-
Hierarchy reversed
- What candidates pick
- AI is a subset of machine learning
- Why it looks right
- The terms are used loosely in the media.
- Why it's wrong
- Machine learning sits inside AI, and deep learning sits inside machine learning; the reverse ordering is false.
Clue to watch for Any statement using 'subset of'.
-
Scanned text
- What candidates pick
- Language AI
- Why it looks right
- The goal is to read words.
- Why it's wrong
- A scanned page is an image until optical character recognition extracts the characters, so the first step is vision.
Clue to watch for Mentions of photos, scans or PDFs of paper forms.
-
Learning without data
- What candidates pick
- Deep learning needs less data
- Why it looks right
- Automatic feature learning sounds efficient.
- Why it's wrong
- Deep networks have many parameters and generally need more labelled examples than classic ML to generalise well.
Clue to watch for A small dataset in the scenario.
-
Generation versus labelling
- What candidates pick
- Generative AI for sentiment tagging
- Why it looks right
- Generative AI is the topical answer.
- Why it's wrong
- Assigning a positive or negative label is a classification task; nothing new is generated.
Clue to watch for Output described as a category or score.
Answer framework
Use these steps when a Domain 1 question describes a business use case and asks what kind of AI it is.
- Identify the input the system receives: rows of numbers, free text, images, video frames or recorded audio.
- Identify the output it produces: a label, a number, a detected object, a transcript, or newly generated content; a churn flag is a label, while next month's sales figure is a number.
- Map input plus output to a task family such as language, vision, speech, forecasting or anomaly detection.
- Decide whether learning from data is involved or whether fixed rules would do the job, which separates ML from rule-based AI; a hand-written tax calculator is rules, while a spam filter trained on examples is ML.
- Check whether the scenario needs new content creation; only then is generative AI the right label, otherwise stay with predictive ML.
Practice question
Want more like this? The 1Z0-1122-26 practice test has questions for every domain.
Checklist
0 of 10
Tick each one you can do without notes. Your ticks stay in this browser.
Study strategy
Budget a short block for this section, roughly three to four hours, and treat it as vocabulary practice rather than technical study. Start with the Oracle University AI Foundations course videos on AI basics; they are free on MyLearn and set the exact framing the questions use. Then build a one-page sheet with the AI, ML, DL and generative AI hierarchy drawn as nested circles, and add two examples of each data type next to it.
Next, practise classification drills. Write ten everyday use cases, such as call transcription, receipt scanning, demand forecasting and product recommendation, and for each one note the input data, the output and the task family. This habit pays off again in Domains 5 and 7, where the same reasoning picks the OCI service.
Finish with a set of practice questions and review every miss by asking whether you misread the input, the output or the hierarchy. Because the items are short, aim for about thirty seconds each so you bank time for the longer OCI scenarios later in the paper.
FAQ
How much of the exam does Intro to AI Foundations cover?
Oracle does not publish section weights for 1Z0-1122-26. Secondary study sites estimate this section as one of the smaller ones, around four of the 40 questions, so plan a short revision block for it.
Is this section technical?
Not very. It tests definitions, the AI, ML and DL hierarchy, and matching use cases to data types. No coding or mathematics is required.
Do I need to know about AGI?
Only enough to know it is a hypothetical goal and that current systems, including LLMs, are narrow AI.
Where does generative AI fit?
Oracle's course places generative AI inside deep learning, which is itself inside machine learning and AI.
Which version of the exam does this page cover?
The 2026 release, 1Z0-1122-26. Secondary sources say it replaced 1Z0-1122-25 during 2026, but confirm the code in Oracle MyLearn before booking.
What is the fastest way to revise this section?
Draw the hierarchy, list examples of each data type, and practise mapping use cases to task families until it takes seconds.
Other Artificial Intelligence & Machine Learning certifications
Where the same subject shows up in other exams. Cards marked "Same subject" open the matching domain.
Test your Domain 1 knowledge
You have reviewed AI basics, data types and the AI, ML and DL hierarchy. Check that you can classify use cases quickly under exam timing.
Keep going
- Training Training Built on your Weak areas
- Practice test Questions by domain, with explanations
- Mock exam Full length, timed like the real exam
- Readiness test Find your weakest domain in minutes
- Cheat sheet Key terms and formulas on one page
- Ask the 1Z0-1122-26 AI tutor Get any concept explained your way