Starting an AI or ML project requires careful alignment across engineering, product, legal, and ethics stakeholders. From aligning on goals and defining data requirements to establishing working groups and identifying bias risks, AI project kickoffs have a distinct professional vocabulary. This exercise covers the collocations used when initiating AI initiatives and communicating their scope and constraints to stakeholders.
0 / 5 completed
1 / 5
The data science lead was asked to ___ the AI project goals with product management and legal before writing the technical specification.
Align the project goals is the standard cross-functional kickoff collocation — before beginning an AI project, teams must 'align' on objectives, constraints, and success criteria. 'Confirm' implies checking something already agreed; 'agree' is the outcome of alignment; 'discuss' is too informal. 'Align' implies active negotiation across stakeholders to achieve shared direction.
2 / 5
The ML engineer recommended they ___ a baseline model first before committing to a more complex transformer-based architecture.
Establish a baseline model is the standard ML project methodology collocation — baselines are 'established' as reference points against which improvements are measured. 'Build' is also common; 'create' and 'set' are less idiomatic. 'Establish a baseline' is the canonical phrase in machine learning and data science for defining the minimum performance standard before experimentation.
3 / 5
The project manager insisted that the team ___ the data requirements and access permissions before the sprint planning meeting.
Define the data requirements is the standard AI project initiation collocation — 'defining' requirements means formally stating what data is needed, in what format, and under what access conditions. 'Specify' is also correct; 'clarify' implies resolving ambiguity; 'confirm' implies checking existing decisions. 'Define' is the primary verb for formally establishing requirements before a project begins.
4 / 5
The ethics board asked the team to ___ potential bias risks in the training dataset before proceeding with model development.
Identify bias risks is the standard responsible AI and project kickoff collocation — risk identification is a formal, deliberate process that 'identifies' specific threats. 'Find' is informal; 'flag' is used for urgent one-off signals; 'surface' implies making existing findings visible. 'Identify' is the canonical term in AI risk assessment frameworks for systematically discovering potential failure modes.
5 / 5
The head of data science chose to ___ a cross-functional AI working group to coordinate the company's first LLM integration.
Establish a working group is the formal governance collocation — working groups are 'established' as deliberate, mandated bodies with defined scope and membership. 'Set up' is informal; 'form' is also used; 'create' is neutral. 'Establish' implies that the working group has formal authority and a defined mandate from leadership, which is the appropriate framing for AI governance.