Summary Task 1 - Outsourced Intelligence

OUTSOURCED INTELLIGENCE
WHAT TO EXPECT FROM ARTIFICIAL INTELLIGENCE (AGRAWAL)

 Cognitive science – analogy between mind & program, try to combine these 2 & use
human thinking processes to make computers work, try to make processes that
resemble neurons in the brain





 CRUM (computational representational understanding of the mind) – mind has
mental representations & these are analogous to data structures in a computer
 Want to explore thinking in humans & try to compare it with AI
 Human thinking processes are not entirely understood – makes it difficult to apply
it to AI
 AI can make sth that has been expensive abundant & cheap
 Main task: prediction – to take information you have & generate new information
 Anticipating what will happen in the future
 The anatomy of a task
 Actions are shaped by underlying conditions & resolution of uncertainty to
lead to outcomes
 Use judgement & prediction
 AI has reduced cost of prediction & made data more valuable
 Judgement – ability to make considered decisions
 Understanding the impact different actions will have on outcomes in light of
predictions
 Tasks with less need for human judgement – more easily automated
 Understanding the level of judgement necessary to pursue actions
 Ways in which AI may affect workplace
 Where whole decisions can be clearly defined with algorithm  AI
 Takes longer in areas where judgement can’t be easily described

MACHINE LEARNING

 Most useful in environments with high degree of complexity
 Costs of machine learning-based predictions dramatically reduced in recent years
 Example: identifying an apple – machine references info from past images of apples to
predict whether an unidentified new image contains an apple

EMPLOYING PREDICTION MACHINES

 Machines that can generate reliable predictions & rely on those predictions to
determine what to do next

,  In some contexts: role of human judgement will become limited (e.g., business related
language translation)
 In other contexts: increased value for human-led judgement tasks (e.g., automated
replies to email proposed BUT human has to judge which response is most
appropriate)
 Medicine – treatment & care will still rely on human judgement

THE MANAGERIAL CHALLENGE

 As AI improves, predictions by machines will increasingly take place of prediction by
humans
 3 insights
 Prediction is not the same as automation
 Automation – tasks are made up of data (collection), prediction, judgement,
action
 Machine learning – only involves prediction
 The most valuable workforce skills involve judgement
 Judgement is complementary to prediction
 if price of prediction falls due to AI advancement, judgment will be in higher
demand
 Managing may require a new set of talents & expertise
 Role will increasingly involve determining how best to apply AI

KEY CHALLENGES

 Shifting training of employees from focus on prediction-related skills to judgement-
related skills
 Assessing rate & direction of adoption of AI technologies to properly time workforce
training
 Developing management processes that build most effective teams of judgement-
focused humans & prediction-focused AI agents


ARTIFICIAL INTELLIGENCE IN PSYCHOLOGICAL PRACTICE: CURRENT & FUTURE APPLICATIONS &
IMPLICATIONS (LUXTON)

WHAT IS AI?

 Technology designed to perform activities that normally require human intelligence
 Important branches
 Machine learning – ability of computers to learn without being explicitly programmed
 Artificial neural networks – mathematical, computational, technological models that
mimic logic & learning functions of neurons in the brain
 Natural language processing – how computers process human natural languages
 Turing test – method for judging intelligence of machines