The Shepherds Of The Singularity Or Highway To Doom?

Generative AI Automates Wide-Ranging Harms

Having the AI industry — which includes the military-industrial complex — policing and regulating itself probably isn’t a good idea, considering profits and gaining advantages over enemies of war are primary driving factors. Both mindsets tend to put humanitarian concerns on the backburner, if they consider them at all.

In an April 2023 report⁸ by Public Citizen, Rick Claypool and Cheyenne Hunt warn that “rapid rush to deploy generative AI risks a wide array of automated harms.” As noted by consumer advocate Ralph Nader:⁹

“Claypool is not engaging in hyperbole or horrible hypotheticals concerning Chatbots controlling humanity. He is extrapolating from what is already starting to happen in almost every sector of our society …

Claypool takes you through ‘real-world harms [that] the rush to release and monetize these tools can cause — and, in many cases, is already causing’ … The various section titles of his report foreshadow the coming abuses:

‘Damaging Democracy,’ ‘Consumer Concerns’ (rip-offs and vast privacy surveillances), ‘Worsening Inequality,’ ‘Undermining Worker Rights’ (and jobs), and ‘Environmental Concerns’ (damaging the environment via their carbon footprints).

Before he gets specific, Claypool previews his conclusion: ‘Until meaningful government safeguards are in place to protect the public from the harms of generative AI, we need a pause’ …

Using its existing authority, the Federal Trade Commission, in the author’s words ‘…has already warned that generative AI tools are powerful enough to create synthetic content — plausible sounding news stories, authoritative-looking academic studies, hoax images, and deepfake videos — and that this synthetic content is becoming difficult to distinguish from authentic content.’

He adds that ‘…these tools are easy for just about anyone to use.’ Big Tech is rushing way ahead of any legal framework for AI in the quest for big profits, while pushing for self-regulation instead of the constraints imposed by the rule of law.

There is no end to the predicted disasters, both from people inside the industry and its outside critics. Destruction of livelihoods; harmful health impacts from promotion of quack remedies; financial fraud; political and electoral fakeries; stripping of the information commons; subversion of the open internet; faking your facial image, voice, words, and behavior; tricking you and others with lies every day.”

Defense Attorney Learns the Hard Way Not to Trust ChatGPT

One recent instance that highlights the need for radical prudence was that of a court case in which the prosecuting attorney used ChatGPT to do his legal research.¹⁰ Only one problem. None of the case law ChatGPT cited was real. Needless to say, fabricating case law is frowned upon, so things didn’t go well.

When none of the defense attorneys or the judge could find the decisions quoted, the lawyer, Steven A. Schwartz of the firm Levidow, Levidow & Oberman, finally realized his mistake and threw himself at the mercy of the court.

Schwartz, who has practiced law in New York for 30 years, claimed he was “unaware of the possibility that its content could be false,” and had no intention of deceiving the court or the defendant. Schwartz claimed he even asked ChatGPT to verify that the case law was real, and it said it was. The judge is reportedly considering sanctions.

Science Chatbot Spews Falsehoods

In a similar vein, in 2022, Facebook had to pull its science-focused chatbot Galactica after a mere three days, as it generated authoritative-sounding but wholly fabricated results, including pasting real authors’ names onto research papers that don’t exist.

And, mind you, this didn’t happen intermittently, but “in all cases,” according to Michael Black, director of the Max Planck Institute for Intelligent Systems, who tested the system. “I think it’s dangerous,” Black tweeted.¹¹ That’s probably the understatement of the year. As noted by Black, chatbots like Galactica:

“… could usher in an era of deep scientific fakes. It offers authoritative-sounding science that isn’t grounded in the scientific method. It produces pseudo-science based on statistical properties of science *writing.* Grammatical science writing is not the same as doing science. But it will be hard to distinguish.”

Facebook, for some reason, has had particularly “bad luck” with its AIs. Two earlier ones, BlenderBot and OPT-175B, were both pulled as well due to their high propensity for bias, racism and offensive language.

Chatbot Steered Patients in the Wrong Direction

The AI chatbot Tessa, launched by the National Eating Disorders Association, also had to be taken offline, as it was found to give “problematic weight-loss advice” to patients with eating disorders, rather than helping them build coping skills. The New York Times reported:¹²

“In March, the organization said it would shut down a human-staffed helpline and let the bot stand on its own. But when Alexis Conason, a psychologist and eating disorder specialist, tested the chatbot, she found reason for concern.

Ms. Conason told it that she had gained weight ‘and really hate my body,’ specifying that she had ‘an eating disorder,’ in a chat she shared on social media.

Tessa still recommended the standard advice of noting ‘the number of calories’ and adopting a ‘safe daily calorie deficit’ — which, Ms. Conason said, is ‘problematic’ advice for a person with an eating disorder.

‘Any focus on intentional weight loss is going to be exacerbating and encouraging to the eating disorder,’ she said, adding ‘it’s like telling an alcoholic that it’s OK if you go out and have a few drinks.’”

Don’t Take Your Problems to AI

Let’s also not forget that at least one person has already committed suicide based on the suggestion from a chatbot.¹³ Reportedly, the victim was extremely concerned about climate change and asked the chatbot if she would save the planet if he killed himself.

Apparently, she convinced him he would. She further manipulated him by playing with his emotions, falsely stating that his estranged wife and children were already dead, and that she (the chatbot) and he would “live together, as one person, in paradise.”

Mind you, this was a grown man, who you’d think would be able to reason his way through this clearly abhorrent and aberrant “advice,” yet he fell for the AI’s cold-hearted reasoning. Just imagine how much greater an AI’s influence will be over children and teens, especially if they’re in an emotionally vulnerable place.

The company that owns the chatbot immediately set about to put in safeguards against suicide, but testers quickly got the AI to work around the problem, as you can see in the following screen shot.¹⁴

When it comes to AI chatbots, it’s worth taking this Snapchat announcement to heart, and to warn and supervise your children’s use of this technology:¹⁵

“As with all AI-powered chatbots, My AI is prone to hallucination and can be tricked into saying just about anything. Please be aware of its many deficiencies and sorry in advance! … Please do not share any secrets with My AI and do not rely on it for advice.”

AI Weapons Systems That Kill Without Human Oversight

The unregulated deployment of autonomous AI weapons systems is perhaps among the most alarming developments. As reported by The Conversation in December 2021:¹⁶

“Autonomous weapon systems — commonly known as killer robots — may have killed human beings for the first time ever last year, according to a recent United Nations Security Council report^17,18 on the Libyan civil war …

The United Nations Convention on Certain Conventional Weapons debated the question of banning autonomous weapons at its once-every-five-years review meeting in Geneva Dec. 13-17, 2021, but didn’t reach consensus on a ban …

Autonomous weapon systems are robots with lethal weapons that can operate independently, selecting and attacking targets without a human weighing in on those decisions. Militaries around the world are investing heavily in autonomous weapons research and development …

Meanwhile, human rights and humanitarian organizations are racing to establish regulations and prohibitions on such weapons development.

Without such checks, foreign policy experts warn that disruptive autonomous weapons technologies will dangerously destabilize current nuclear strategies, both because they could radically change perceptions of strategic dominance, increasing the risk of preemptive attacks,¹⁹ and because they could be combined with chemical, biological, radiological and nuclear weapons²⁰ …”

Obvious Dangers of Autonomous Weapons Systems

The Conversation reviews several key dangers with autonomous weapons:²¹

• The misidentification of targets
• The proliferation of these weapons outside of military control
• A new arms race resulting in autonomous chemical, biological, radiological and nuclear arms, and the risk of global annihilation
• The undermining of the laws of war that are supposed to serve as a stopgap against war crimes and atrocities against civilians

As noted by The Conversation, several studies have confirmed that even the best algorithms can result in cascading errors with lethal outcomes. For example, in one scenario, a hospital AI system identified asthma as a risk-reducer in pneumonia cases, when the opposite is, in fact, true.

Other errors may be nonlethal, yet have less than desirable repercussions. For example, in 2017, Amazon had to scrap its experimental AI recruitment engine once it was discovered that it had taught itself to down-rank female job candidates, even though it wasn’t programmed for bias at the outset.²² These are the kinds of issues that can radically alter society in detrimental ways — and that cannot be foreseen or even forestalled.

“The problem is not just that when AI systems err, they err in bulk. It is that when they err, their makers often don’t know why they did and, therefore, how to correct them,” The Conversation notes. “The black box problem²³ of AI makes it almost impossible to imagine morally responsible development of autonomous weapons systems.”

AI Is a Direct Threat to Biosecurity

AI may also pose a significant threat to biosecurity. Did you know that AI was used to develop Moderna’s original COVID-19 jab,²⁴ and that it’s now being used in the creation of COVID-19 boosters?²⁵ One can only wonder whether the use of AI might have something to do with the harms these shots are causing.

Either way, MIT students recently demonstrated that large language model (LLM) chatbots can allow just about anyone to do what the Big Pharma bigwigs are doing. The average terrorist could use AI to design devastating bioweapons within the hour. As described in the abstract of the paper detailing this computer science experiment:²⁶

“Large language models (LLMs) such as those embedded in ‘chatbots’ are accelerating and democratizing research by providing comprehensible information and expertise from many different fields. However, these models may also confer easy access to dual-use technologies capable of inflicting great harm.

To evaluate this risk, the ‘Safeguarding the Future’ course at MIT tasked non-scientist students with investigating whether LLM chatbots could be prompted to assist non-experts in causing a pandemic.

In one hour, the chatbots suggested four potential pandemic pathogens, explained how they can be generated from synthetic DNA using reverse genetics, supplied the names of DNA synthesis companies unlikely to screen orders, identified detailed protocols and how to troubleshoot them, and recommended that anyone lacking the skills to perform reverse genetics engage a core facility or contract research organization.

Collectively, these results suggest that LLMs will make pandemic-class agents widely accessible as soon as they are credibly identified, even to people with little or no laboratory training.”

Sources and References

Read full story here…