Tactical Development of AI (Executive Management) Flashcards
(20 cards)
LLMs?
large language models
raise productivity
ai management system?
tool developed to implement, operate, supervise & improve AI technical capabilities within an organisation
key objectives senior executives should pursue when running ai management systems?
fairness
security
privacy
robustness
transparency
accountability
availability
maintainability
data quality
transparency & explicability
is appropriate degree of ai expertise important in an organisation deploying ai systems?
yes
human resources
hr planning & staffing requirements which are needed to use ai systems should be properly documented
common positions involved in the development & operation of ai systems?
data scientists
specialist supervisors
specialist researches
database engineers
technically orientated customer service staff
requirements to work in a technical capacity on an ai project
uni degree in quantitative discipline & relevant experience
data analysis tools
python, r SAS, SQL
examples of technology required for an AI system?
relational databases
uses an API
running on a GPU computer
uses cloud computing solutions
data visualisation apps
impact assesment?
to determine the effect of an ai system on individuals and larger society
via assessing the consequences of deploying the system
specific aspects of ai impact assessment
consider the context in which it’s used
the specific components of an assessment
staff must be qualified, available and non-conflicted
management make decisions
determine who stakeholders of the ai system are
main technical resources required for ai project
data resources
software tools
hardware
back-up/disaster recovery
steps in the ai system lifecycle
determining & documenting objectives
creating method as to how they can be achieved
prioritising the objectives
ascertaining how this may affect various parts of the development phase
ANN?
artificial neural network
other actions & considerations related to the AI lifecycle
- determining which stages require impact assessments
- choosing tolerances for testing the model & the methodology
- current level of expertise
- approval process at each stage
- rule changes
- ensuring maintainability
- ensuring maintainability
- setting procedures for monitoring & continuous improvement
- engagement with the system stakeholders
technical considerations which go into the development phase of an ai system’s lifecycle
machine learning technique
algorithms
how the model’s trained/tested
hardware needed
non-machine learning model software
whether there are likely security threats
how outputs are presented/distributed
degree of human interaction
compatibility with other systems
when is deciding the type/degree of oversight important?
when an ai system has the potential to impact individuals & societies - not just interacting with other technologies
management must consider what concerning data for ai systems
sources of data
frequency & timeliness of data
different time periods of groupings
how data is collected
qualitative categories of data
quality
how data will be used
how data is pre processed
data used in an ai system could come from
open sources
specialised vendors
organisation’s own customers
anywhere else which can generate business insights
important elements to be shared with interested parties
reasons for using ai
fact that not interacting with a human, but ai
how to effectively interact with the system
technical requirements for using the system
key items from impact assessment
contact info
instructional material
