Please enable JavaScript.
Coggle requires JavaScript to display documents.
The risks to AI are significant but controllable (How to tackle the risks,…
The risks to AI are significant but controllable
New technologies
Improving our lives
Simplifying shopping
Healthcare experiences
Enormous potential
Additional economic output of $13 trillion by 2030
How to tackle the risks
Clarity about the most worrisome scenarios
Using a group of risk-pinpointing managers
Risk-identification has become a well-developed art
The application of company-wide controls
Proper oversight
Strong policies/procedures
Worker training
Managers "in the loop"
Identifying special circumstances
Taking human difficulties into account
Gap analysis
Framework of risk-management
Areas deepened
Areas redefined
Areas extended
Requiring of fallback plans and scenario planning
Some unwanted and serious consequences
Applying to all advanced analytics
Most visible ones
Privacy violations
Discrimination
Accidents
Manipulation of political systems
Consequences not yet known
Most concerning
Loss of human life?
Compromise of national security
Disinformation to a military AI
Risks of AI
Data difficulties
Inadvertently using or revealing sensitive information
Amount of unstructured data has increased
Technology troubles
Technology and process issues
Possible negative influence
For example
Failing to spot trading issues
Security snags
Potential for fraudsters
Exploit collected data by companies
Creating of false identities
Consumer backlash and regulatory repercussions
Models misbehaving
Delivering of biased results
Yielding wrong conclusions
AI lurking in SaaS offerings
Giving rise to hidden vulnerabilities that hackers might exploit
Interaction issues
Challenges in automated systems
Accidents when humans don't pay attention