logo

Intelligence in Our Image Risks of Bias and Errors in Artificial Intelligence

PDF Publication Title:

Intelligence in Our Image Risks of Bias and Errors in Artificial Intelligence ( intelligence-our-image-risks-bias-and-errors-artificial-inte )

Previous Page View | Next Page View | Return to Search List

Text from PDF Page: 035

CHAPTER FOUR Conclusion We have illustrated the challenge of algorithmic disparate impact, why we can expect to see an expansion of algorithm dependence, and what the best options for mitigating future risk might be. The error and bias risk in algorithms and AI will continue as long as artificial agents play increasingly prominent roles in our lives and remain unregulated. Response to unregulated artificial agents tends to be of three broad types: avoiding algorithms altogether, making the underlying algorithms transparent, or auditing the output of algorithms. Avoiding algorithms is probably impossible; few other options are available for making sense of the current deluge of data. Algorithmic transparency requires a more educated public capable of understanding algorithms. But recent advances in deep connectionist learning mean that, even if we could deconstruct an algorithm’s procedure, it may still be too com- plex to make useful sense of that insight. Christian Sandvig’s recent work argues that the last option, the algorithm audit, should be the way forward (Sandvig et al., 2014). Cer- tain audit types ignore the inner workings of artificial agents and judge them according to the fairness of their results. This is akin to how we often judge human agents: by the consequences of their outputs (deci- sions and actions) and not on the content or ingenuity of their code base (thoughts). This option makes the most sense for policymakers and sets the standard for a consequentialist ethics for artificial agents. Regulation is much easier under this framing. Discussions like this one may sometimes anthropomorphize arti- ficial agents: Are machines beginning to think like us, and how can 25

PDF Image | Intelligence in Our Image Risks of Bias and Errors in Artificial Intelligence

intelligence-our-image-risks-bias-and-errors-artificial-inte-035

PDF Search Title:

Intelligence in Our Image Risks of Bias and Errors in Artificial Intelligence

Original File Name Searched:

RAND_RR1744.pdf

DIY PDF Search: Google It | Yahoo | Bing

Cruise Ship Reviews | Luxury Resort | Jet | Yacht | and Travel Tech More Info

Cruising Review Topics and Articles More Info

Software based on Filemaker for the travel industry More Info

The Burgenstock Resort: Reviews on CruisingReview website... More Info

Resort Reviews: World Class resorts... More Info

The Riffelalp Resort: Reviews on CruisingReview website... More Info

CONTACT TEL: 608-238-6001 Email: greg@cruisingreview.com | RSS | AMP