With support from the Minderoo Foundation, the Gradient Institute recently released an analysis of the increasing threat from Artificial Intelligence (AI) in companies and open-source software that companies can use to mitigate the risks.
Bill Simpson Young, the CEO at Gradient Institute and former CSIRO, discusses the risks associated with Artificial Intelligence for business.
What are the risks from Artificial Intelligence we need to be aware of in the business world?
AI employed by companies has many advantages for companies and their clients, like taking actions for customers in a rapid manner and tailoring specifically for the customer. For instance, each time you access a map app on your smartphone, speed, accuracy, and relevancy to you are achieved through AI. AI also plays a role in determining your feed of news and determining whether you’re an interview for a job and whether you’re successful with your loan application.
As we demonstrate in this report, there is overwhelming evidence to show that the application of AI for automated decision-making can result in illegal, inhumane, and inequitable outcomes for individuals through typically non-transparent and unaccountable decisions.
These problems stem from the unjustified trust placed in (or at least trust in) AI. Machines and humans make decisions in different ways. While humans can use the ability to use common sense and to navigate various situations effortlessly, machines do not have a built-in moral judgment. They are only able to perform their best in narrow areas.
What are the things we must be aware of to protect our company?
Companies that run AI systems that could impact the lives of their users should be better at knowing the risks that AI systems pose for their businesses. The report contains an assessment of these new risks that include “failures of legitimacy” (for instance, when how an AI system functions lead to intentionally treating different types of individuals differently, and therefore in violation of anti-discrimination laws), “failures of design” (for example, when an AI system is being created using data that isn’t appropriate for the choices that the AI is expected to make) as well as “failures of execution” (for instance, failing to adequately check the performance that the AI system with time).
What are the best ways to fight it?
It is a practical one. We outline several actions companies can undertake to make sure they responsibly use AI. Many existing strategies can help AI be utilized responsibly, and we’ve brought them together to make it easier for companies to implement these strategies. We have also released software we hope that businesses that use AI will utilize to give more control and oversight for your AI systems. We have named it AI Impact Control Panel software. We have made it accessible as an open-source program so that companies can use it and modify it easily and without difficulty.
What are the most important conclusions of the report?
This report offers guidelines for businesses to minimize the risk of adopting AI systems to make decisions. We discuss how AI systems make different decisions than humans and how these variations create new types of risk for organizations. We present a taxonomy that can assist companies in thinking about AI risks and suggest ways to deal with the issues. We hope that companies will find this document and the software useful and create more accountable AI systems. So, the public and businesses can enjoy all the advantages of AI while staying clear of many of the negative effects on society and individuals that are currently taking place.
More About The Open Source Software
Businesses that employ Artificial Intelligence (AI) will soon have access to open-source software to better manage the effects of the actions of AI systems. AI systems.
The program was developed through Gradient Institute. With support from Minderoo Foundation’s Frontier Technology initiative. The AI Impact Control Panel elicits the motivations and preferences of decision-makers via a graphic user interface. It then converts them to the language of mathematics needed for an AI system.
People who use the software don’t require any technical understanding of AI. Instead, they can establish the objectives and limits to AI. AI system.
Control Panel Control Panel helps ensure the AI system’s operations are in line with the ideals of the society and organization by continually asking the individuals responsible in this AI system what the acceptable ranges for different performance measures are (compared with baselines known) and what the importance is of various goals and the relative importance of various outcomes. The software adapts the choices given to users in time, allowing them to quickly identify their preferences without overburdening them.
Minderoo Foundation:
Created in 2001 by Andrew Forrest and Nicola Forrest 2001, Minderoo Foundation is a modern charitable organization that aims to overcome barriers, innovating and bringing about positive, sustainable transformation. Minderoo Foundation is proudly Australian and has a range of key initiatives from ocean research to the end of slavery to collaboration to fight cancer, establish communities, and strengthen our digital economics.
Gradient Institute:
Gradient Institute is an independent non-profit research institute that works to bring ethics, accountability, and transparency into AI systems by creating new algorithms, providing training to companies running AI systems, and offering technical advice to assist with AI policy creation.
