Costs and Benefits
AI is a double-sided coin...
AI and Electricity
According to a recent report by Bloomberg, artificial intelligence uses so much power that it is putting a strain on other aspects of the US power grid. The enormous energy requirements of the data centers that make AI possible have begun causing an issue called bad harmonics, or the interruption of normal electric current with unpredictable fluctuations in voltage. This increases the risks of appliance damages and household fires in the surrounding areas. Currently, those living in close proximity to data centers are most impacted; over 75% of all unusual power readings in the US are within 50 miles of high data center activity. Furthermore, as AI continues to expand, this problem will as well. Data centers can be constructed far more rapidly than the power grid can be adequately adjusted to prepare for them. Without significant investment in the US power grid – an estimated $3.1 trillion is necessary before 2030 to both meet consumption needs and help transition to renewable energy – the issues that come with bad harmonics will only worsen.
Takeaway: Artificial intelligence places an enormous strain on our resources. If we are to continue at this current rate of expansion, serious consideration must be given to how we will meet the demand that AI requires while protecting both households and the environment. –Grace Ogden
To read more about data centers’ strain on the power grid, please visit Data Center Dynamics.
An AI Labyrinth
Just recently, cybersecurity company Cloudflare announced a new system to thwart artificial intelligence bots, known as crawlers, attempting to scrape unauthorized data from websites to train other models. Within the website being protected, this AI Labyrinth method creates links to AI-generated web pages that look convincing to other bots but contain no real data, causing crawlers to waste energy, resources, and time exploring them. Furthermore, AI Labyrinth can help detect crawler activity, as real humans would not traverse multiple links into web pages of nonsensical AI-generated content. Therefore, anyone viewing these pages is most likely a bot. This allows Cloudflare to identify and divert crawlers without alerting the attacker and allowing them to change their approach.
Takeaway: Sometimes, it’s effective to fight fire with fire, and AI Labyrinth is a prime example. As artificial intelligence continues to evolve, it will certainly have malicious uses, but it can also be employed to protect against those same attacks. –Grace Ogden
To read more about AI Labyrinth, please visit Cloudflare.
New Jersey Anti-Deepfake Law
New Jersey passed a law criminalizing the creation and distribution of deepfake media. This comes after increasingly common incidents of deepfake pornography generated by students in high and middle schools, prompted by easy accessibility to deepfake generators and a lack of regulation to provide grounds for prosecution. The criminalization of deepfakes specifically is an important step to regulating the technology and providing prosecutors with legal avenues to punish bad actors using deepfakes. The law defines deepfakes using clear, bipartisan language, setting a standard for other states to follow suit in the US and abroad.
Takeaway: The passage of a law criminalizing deepfake production provides legal processes for prosecution of malicious users as well as a precedent for other states to follow suit. This is an important step in regulating AI use for the benefit of the public. –Natalie Sherman
To read more about the law passed in New Jersey, please visit AP News.