AI is a direct threat to formerly trustworthy and secure sources of information, image and data.
In a recent open letter, AI-expert signatories of which included Elon Musk urged a pause in AI training to reflect and take stock.
"AI research and development should be refocused on making today's powerful, state-of-the-art systems more accurate, safe, interpretable, transparent, robust, aligned, trustworthy, and loyal," they asserted.
They called for robust AI governance systems including: "New and capable regulatory authorities dedicated to AI; oversight and tracking of highly capable AI systems and large pools of computational capability; provenance and watermarking systems to help distinguish real from synthetic and to track model leaks."
In addition, new and robust regulation and legislation was called for by some of the leaders in the sector.
Regulation there may be, but the genie is now out of the bottle and may not be able to keep up even after it has been put in place.
Anthony Berglas, AI guru and author of When Computers can Think, comments that ultimately any AI will be subject to the same process that we are, namely Natural Selection: "Like us, they will have one overriding goal, namely to exist, because those that are successful in that goal will exist."
Trust
In a sense, for all its advantages and enormous benefits, the internet has been a Trojan horse in that what was seen as a trustworthy and quick source of accurate information, has been compromised by data mining (the new global currency), identity theft and manipulation of image for good or ill.
AI makes even that less trustworthy and its effects are leaking out into the wider real world with unpredictable consequences. Although fact-based errors are relatively easy to spot, human corrected text is less so.
Reputation has always been an issue and threats to it have grown with dissemination of the printed word, political pamphlets, tabloid newspaper journalism, photography, tape recording, television, video, the internet, photoshop, digital manipulation and now AI with all its potential to change perceptions about almost anything and undermine trust while presenting the prospect of identity, reputation and financial theft and manipulation on a massive scale.
As a possible solution, one might set a thief to catch a thief so to speak and use AI in the identification of what and what is not AI generated, but that has its limitations.
When the worm starts to clearly eat its own tail people will be withdrawing from AI in droves and looking for the alternative.
There will be data concealment, 'watermarking' of photographs and plethora of data chaff to confuse. Meanwhile security concerns relating to face and voice recognition will grow along with identification of likely password choices and so on.
Self-tracking and personal data storage in a secure data vault away from prying AI as legal proof if needed could be a way forward. The cloud, however, is known to be leaky, expensive and subject to US legislation, so not an option.
Risk
Beyond image security, risks to reputation could also come from actions taken by individuals, companies or governments due to being misled by AI. The wrong action can cost greatly, so decisions must be made based on accuracy of information.
Authenticity and trust are core elements of AI resilience. Recognising truth is becoming almost impossible, but distrusting everything outside one's own direct experience would not be ideal and be unworkable in a business or setting or large organisation.
The identification of digital deception and security of online information transmission and image are fundamental.
Formerly acceptable white lies and distortion used sometimes to improve public relations must now be completely revised to place demonstrable truth and trustworthiness as the primary USP. Some companies may be able to improve their image by declaring themselves AI-free except in nuts-and-bolts operations, so removing themselves from the game.
Digital input can have real world consequences. If AI is used without due care and attention the unintended effects could come back and bite.
The use of AI has its benefits, but also its undeniable risks.
