The intended outcome of stage 2 training depends entirely on the field you’re asking about. This term appears across driving instructor qualification, AI model development, and medical education, each with distinct goals and standards. Here’s what stage 2 training is designed to achieve in the most common contexts.
Driving Instructor (ADI) Part 2 Test
If you’re training to become an Approved Driving Instructor in the UK, stage 2 (officially called the ADI part 2 test) is designed to prove you can drive at an expert level. The intended outcome is demonstrating that your own driving ability far exceeds that of a typical driver, because you’ll eventually need to teach others while managing a vehicle.
The test evaluates five areas: an eyesight check (reading a number plate from 27 metres), vehicle safety questions, general driving ability, manoeuvres, and independent driving. The eyesight and safety questions are pass/fail basics, but the driving portion is where the bar sits noticeably higher than a standard driving test.
For general driving ability, you need to show expert handling of the controls, correct road procedure, anticipation of other road users’ actions, and sound judgement of distance, speed, and timing. You’re also assessed on consideration for other road users and environmentally friendly driving. The examiner wants to see someone who doesn’t just follow the rules but reads the road with the kind of awareness a professional instructor needs.
The manoeuvres section requires you to complete two exercises chosen from parallel parking, reversing into or out of a parking bay, or pulling up on the right and reversing. You’ll also drive independently for about 20 minutes, following either a sat nav or traffic signs. The intended outcome across all of this is a level of driving competence that would give a learner driver full confidence in their instructor’s skill behind the wheel.
AI and Large Language Model Development
In AI development, “stage 2 training” typically refers to one step in the pipeline used to turn a raw language model into a useful assistant. The standard pipeline has three stages: supervised fine-tuning (SFT), reward modelling, and reinforcement learning. Stage 2 in this sequence is reward modelling.
The intended outcome of reward modelling is building a separate model that can score the quality of the AI’s responses. During this stage, human reviewers rank different outputs, and those rankings train a reward model to predict which answers humans prefer. This reward model doesn’t change how the AI writes directly. Instead, it creates a scoring system that the third stage (reinforcement learning) uses to nudge the AI toward better, more helpful, and safer responses.
Some teams describe their pipeline differently, with stage 2 referring to supervised fine-tuning itself. In that framing, the intended outcome is teaching the base model to follow instructions and respond in a conversational format. Research on two-stage fine-tuning approaches shows that format information (how to structure a response) tends to be learned separately from content knowledge. The model learns to behave like an assistant: answering questions directly, following formatting requests, and staying on topic rather than generating raw text the way it did during initial pre-training.
Medical Specialty Training
In medical education, particularly under the milestone framework used by residency programs, Level 2 (sometimes called stage 2) describes an “advanced beginner” who can perform some clinical tasks with limited autonomy. The intended outcome is moving beyond the novice phase into a trainee who recognizes common patterns and can carry out straightforward procedures, though still under close supervision.
This framework is criterion-based rather than tied to a specific year of training. A resident reaches Level 2 by demonstrating competence in defined subskills, not simply by completing a set number of months. The progression continues through Level 3 (performing common tasks independently), Level 4 (the graduation target for most programs), and Level 5 (exceeding expectations). The purpose of Level 2 is establishing a foundation of clinical reasoning and basic procedural skill that supports the growing independence expected at later stages.
Military Advanced Individual Training
In U.S. military training, the second major phase after Basic Combat Training is Advanced Individual Training (AIT), which functions as stage 2 of a soldier’s development. The intended outcome is job-specific expertise. Where basic training builds general fitness, discipline, and foundational combat skills, AIT produces soldiers who are proficient in their assigned career field.
Depending on the military occupational specialty, this means becoming competent in weapon systems, vehicles, tactical support equipment, or combat operations involving specific platforms like tracked vehicles. For infantry roles, AIT covers small arms, anti-armor weapons, indirect fire weapons, vehicle operation and maintenance, and land reconnaissance. The goal is a soldier who arrives at their first unit ready to perform their role with minimal additional instruction.