SUMMARY
Beyond sensationalist AI doomsday scenarios, this talk refocuses on a more immediate concern: the risks posed everyday algorithms that shape our world. It highlights the very immediate and real threat posed by current algorithms in social media, policymaking, law enforcement, and fraud detection, capable of creating subtle but nonetheless nasty dystopias. This talk suggests that if we don’t design algorithms well today, Terminator scenarios are the least of our worries tomorrow.
Delving into the intricate landscape of algorithmic influence and AI impact , this session unravels the nuanced challenges product teams face. By navigating through practical strategies like transparency, explainability, and human intervention, this talk explores how to wield algorithms to our advantage without inadvertently steering society toward discrimination, exploitation, and oppression.
This is not just a theoretical exercise; it's a call to action for all levels of IT professionals to proactively shape the algorithms that shape our world.
DESCRIPTION
This talk takes participants on a journey from existential risks, including AI doomsday scenarios, to the more imminent dangers posed by the algorithms shaping our daily lives.
I’ll highlight:
The imperative to thoughtfully design and deploy all decision-making algorithms - AI or not.
The existing principles—transparency, explainability, human intervention—that must guide the design of algorithmic systems.
How the talk unfolds:
The talk unfolds with an entertaining exploration of existential risks, touching on everything from aliens, Gray Goo, super intelligent AI to the Great Filter.
I'll then pivot to why these scenarios, while fascinating, divert our attention from the far more immediate risks embedded in the algorithm we currently use. From policy-making to fraud detection, these algorithms, often unnoticed, are already shaping our society, and, if badly done, can have very ‘hasty’ consequences.
I will briefly outline why the full spectrum from super-intelligent, to general or weak AI to our most basic algorithms need our considerations.
Drawing from industry examples, I'll showcase the good, the bad, and the ugly in algorithm implementation.
I will conclude with actionable recommendations: I'll advocate for best practices, including data access rights, transparency, and the importance of options for human intervention in AI driven systems.
These solutions empower us to address challenges and risks today while preparing for the future emergence of stronger AI.
TAKEAWAY
Participants in my talk will
Recognize the critical importance of assessing risks in algorithmically driven systems beyond the realm of (strong) 'AI.
Understand how to integrate impact and risk considerations into the design process
Gain insights into practical solutions to designing systems to proactively mitigate unintended or detrimental consequences.