|▲ IEEE 인공지능표준의장 투피 살리바와 박영숙|
Global AI Standards
Global AI Standards SIG
Toufi Saliba – Global Chair Ai Standards for all humankind , IEEE TEMS
Problem: Perhaps the biggest existential threat to our existence. While many are looking to prevent AI from getting rid of humanity, we look to secure AI to prevent some human from using it to attack other humans. Currently in the world today, not a single AI is being built preventing an attack from within. Attacks can be as simple as repurposing. The best way to think of thousands of scenarios, think you are in control of some AI processes in any territory, now think you handed that control to your worst enemy. What can they do, using the power it has? At this point any reader would think but “you won’t do that” .. well, your worst enemy has the ability to build and license that same AI perhaps in different territory today. Visionaries like Elon Musk seem to get a gist of what the problem we are facing but without knowing what can be done to prevent it technologically. Sure, you may think regulating over regulating etc. That thinking may work for a short period of time in single jurisdiction.
Solution Requirements: AI that can not be re purposed, can not be attacked from within to be used against the people without the consent of the people. Might sound not possible. We challenge that and don’t think the life of humanity and future generations should suffer from skeptics who prefer keeping the status quo or not examining truly if these requirements can be achieved. This is our raison d’être, one approach might be utilizing Autonomous Decentralized Governance, one form of it is User Centric Governance. The end results may very well be software development standards, communication standards that will enable these requirements to be met. At that point any software not compliant with the standards will likely lose its ability from being adopted.
We are launching the Global AI Standards and of you are reading this, we are asking for your help. Together we can make this a reality. Reach out, spread the word and/or join the movement.
Crawl – Walk – Run
During my term I plan to gather as many ideas, individuals, institutions and nations as possible to work together. The individual members are likely to be spread out across the globe and in their areas cover as many technical events as possible spreading the word to get as much help as we can. There will be open source software development initiatives for folks to join along with non-technical initiatives for folks to partake in this mission. There will be centers taking on lead role in providing intensive training for qualified practitioners and academics with the hope to work cooperatively towards solutions. We know the problem exists. We know it’s not reversible and can not be fixed after the fact and we know we can do a lot of things to preventatively fix it and not correctively, not by force me not by regulations but more so by voluntary standardization.
About Toufi Saliba
As well as serving as the founding Global Chair Ai Standards for all humankind at IEEE TEMS, Toufi Saliba is also the CEO at PrivacyShell and former Chair of the ACM Practitioner Board Conference Committee.
Toufi’s background is mainly in Machine Learning, Autonomous Decentralized Governance, Distributed Computing, and Cryptography.
He’s TODA/IP co-author. He has authored and co-authored several algorithms, protocols, and patents. Toufi’s companies have had several exits for software that he built from the ground up, some ended up at Google, HP and Intel. He sits on multiple Silicon Valley start-up boards, and involved in several foundations including President at the Decentralized AI Alliance. He also runs a start-up incubator and InfoSec advisory called PrivacyShell Corp. PrivacyShell’s portfolio founders have an aggregate of over $27B in exits. For the last 2 decades, Toufi’s #1 goal was enhancing technology to help achieve global prosperity while building profitable businesses and enablements from within. He feels confident we are on track and now works on increasing the security by design for AI to prevent attacks from within and ensures it continues to service all humankind for as long as possible.