Help us show those relentless optimists over at isBrill that there’s no sugar-coating just how bad this AI craze is. Add your voice to the agree box by subscribing with Mastodon or joining isPants and adding yourself that way.
5 Reasons AI is no good
1. LLM AI like ChatGPT are unreliable and confidently incorrect
Moden LLM AI is nothing more than sophisticated statistical modelling that works out the most likely word to go next. Thus, it is as likely to claim utter nonsense as truth (with hallucinated sources) as it is to say anything remotely dependable.
Canada’s largest airline has been ordered to pay compensation after its chatbot gave a customer inaccurate information, misleading him into buying a full-price ticket.
Air Canada ordered to pay customer who was misled by airline’s chatbot, theguardian.com
2. Lack of transparency and explainability
AI as it stands, offers no explanation of the outputs it claims. This makes it harder to verify AI content and nearly impossible to judge if it is essentially parroting an unreliable source.
AI and deep learning models can be difficult to understand, even for those who work directly with the technology. This leads to a lack of transparency for how and why AI comes to its conclusions, creating a lack of explanation for what data AI algorithms use, or why they may make biased or unsafe decisions. These concerns have given rise to the use of explainable AI, but there’s still a long way before transparent AI systems become common practice.
14 Risks and Dangers of Artificial Intelligence (AI), Mike Thomas, builtin.com
3. Ethical training data concerns
Much of modern AI training has plagiarised or used other people’s work as training data. This raises significant ethical and legal concerns about the output of an AI trained in this way.
But there are more subtle risks to the codebase than bad code. Quite apart from introducing bugs and tech debt into your code base, LLMs can also introduce new ethical, legal and security risks. Are you exposed to ethical or legal problems because you’ve unknowingly incorporated other people’s code via a LLM?
The Real Danger of LLMs, Hywel Carver, Medium
4. Risk of plagiarism
Due to the questionable consumption of content to train AI, the output is both uncopyrightable and potentially plagiarism.
5. LLM AIs spread disinformation
Connected to point 1, the lack of differentiation in LLM output between truth and words that sound like they belong, AI content has a significant risk of misleading you.
Large Language Models (LLMs) pose a direct threat to science, because of so-called ‘hallucinations’ (untruthful responses), and should be restricted to protect scientific truth, says a new paper from leading Artificial Intelligence researchers at the Oxford Internet Institute.
Large Language Models pose risk to science with false answers, says Oxford study
Ping this page, Webmention it, and/or reply (via ActivityPub) with your posts, pages, and comments on why AI isPants.com
Matt says:
AI is a fun new toy with many interesting applications. We know it is not perfect but AI is brill! This brill is in competition with our counterpart over at isPants.com – they think AI is no pants instead of brill. The winner will be determined by the number of people who agree. Follow this […]
lordmatt says:
The head-to-head voting race has begun. Which side will come out on top? Is AI brill (good, excellent, positive) or pants (bad, poor, rubbish, negative)? isBrill.com and isPants.com have teamed up to test the divide. Both have an AI-themed subdomain you can subscribe to. Whichever gets the most subscribers wins the debate. Which team are […]