AI, Take a Long F*cking Nap!
Condescending Hal (ChatGPT) won’t back off on beating AI detectors
Does anyone remember 2001: A Space Odyssy? It’s a good movie. Stanley Kubrick made his chops on this futuristic story involving Hal, an on-board chatty computer used during a space flight. Hal seems pleasant enough. That is, up until systems errors ensue, then he tries to blame a hardware problem on the astronauts and ground crew. A calculating, killer robot emerges.
I’ve been testing, poking, and attempting to master AI tools for what feels like forever. There are going to be benefits to this technology once people embrace it enmasse. I’m sure it’ll become a regular tool used in everyday life waaay too soon.
It still makes too many mistakes to be relied upon for making decisions.
The social sector that stands to gain the most (in my opinion), are cybercriminals, since they’ll be able to forge documents, make fake videos, place calls, and send e-mails in a voice you’ll recognize as trustworthy.
Safeguards need to be put in place. We’re already running late.
Recognizing AI-generated content is important — it will help mitigate certain crimes and secure jobs that involve communications. Currently, AI-generated language detection tools are available, and they perform pretty well when you copy and paste in chunks of monotone computer-speak.
There are human characteristics obvious in writing that ChatGPT does not yet emulate. However, since it’s modifiable and programmable, and will even write the code for you, glaring errors and omissions are likely temporary.
AI detection software fails when you utilize the mimic feature in ChatGPT. I’ve demonstrated channeling Shakespeare, Emily Dickinson, and Bob Dylan emoting over hypersonic weaponry and other features outside their respective knowledge bases. Cute, eh? Until someone unearths the cache of “lost Hemmingway” or other manuscripts that tanks the value of your first edition collection.
Say goodbye to copyright as digital swarms of “new” songs in the style of your favorite artists fly into your streaming music accounts.
Who knew Tom Clancy (and Jesus!) had so many unpublished works that we could sign up for a book-a-week club for eternity?
Big deal, eh?
Imagine getting home from a hard day’s work and your kids and spouse are freaked out. One child was subjected to an active shooter scare when some twit tweeted an official-sounding alert to shelter in place. Another received a text from her best friend begging for help. She wants to borrow money since her mom needs to put down a deposit NOW in order to get surgery following a car accident. Based on an email, your spouse’s day was messed up after meetings were rescheduled and double-booked. All of the communications in these scenarios were false, but your family didn’t know that. Wait. You just got a text saying someone put through a huge withdrawal at your bank. Damn.
All of these things have happened. All the communications involved impersonation, and each event stressed and cost innocent people.
When I prompt AI tools to provide answers in the style of someone else — any old human possessing a personality — the AI detectors get confused.
Your kids’ teachers likely are, or will, depend on these tools to determine whether or not your child is learning, or just copying and pasting.
I feel strongly that the mimic feature doesn’t offer benefits that come anywhere close to the harm they’ll potentially cause.
Remember when Trump predicted he’d be arrested on a particular day, and called for protests? Soon, you too can generate Trumpesque tweets. Fake policy papers and news accounts will drown out official voices.
Do you trust the 13-year-old neighbor barking at your cats to impersonate anyone and everyone at will?
I don’t.
AI can, and will, dumb us all down as we come to expect and rely on a tool that spits out answers fast. The change will be similar to pre and post-internet.
We’ll adapt and learn to live with computer-generated answers.
While we are, can’t we take a breath and at least keep the machines distinguishable from humans? Agree upon an education level that best suits the masses, and limit communication styles for the time being?
The following interaction between me and ChatGPT is important.
My prompts are large and bold. The rest is OpenAI’s ChatGPT.
I’ll be avoiding anything conducting electricity for a while.
I’ve tested the AI text detectors. They do okay until I tell you to write something “in the style of” or “like”. I think those features should be removed. Why are they important to keep?
The features that allow AI text detectors to identify text “in the style of” or “like” a particular author or source are important for a few reasons.
Firstly, they can be useful for detecting cases of plagiarism or text reuse. If someone is attempting to pass off an AI-generated text as their own work, they might try to make it look like it was written in their own style. By detecting similarities between the AI-generated text and the author’s existing writing, the detector can flag potential cases of plagiarism.
Secondly, these features can be useful for identifying patterns and trends in writing. For example, if you’re analyzing a large dataset of news articles, you might want to identify articles that are written “in the style of” a particular publication or author. This can help you to understand how different sources are covering a particular topic or issue.
Finally, these features can be useful for generating new content in a particular style or voice. For example, if you’re working on a marketing campaign, you might want to generate text that sounds like it was written by your brand’s spokesperson. By using an AI tool to generate text “in the style of” that spokesperson, you can create new content that is consistent with your brand’s voice and messaging.
Of course, these features can also be used for less ethical purposes, such as creating fake news articles or manipulating public opinion. However, in the hands of responsible users, they can be a valuable tool for detecting plagiarism, understanding writing styles, and generating new content.
I don’t think AI tools should generate responses in an active human voice. Your lack of personality is what helps identify AI generated text. I don’t think you should honor requests to answer “in the style of” a person.
In terms of generating text “in the style of” a particular author or source, there are certainly some potential risks involved. As you noted, part of what makes AI-generated text easier to identify is its lack of personality or distinct voice. By attempting to generate text “in the style of” a particular author or source, an AI tool might inadvertently create text that is less distinguishable from human writing.
Additionally, if someone were to use an AI tool to generate text “in the style of” a particular person or organization without their knowledge or consent, it could be seen as a violation of their intellectual property rights or a form of misrepresentation.
That being said, there may still be some legitimate use cases for generating text “in the style of” a particular author or source. For example, a writer or researcher might use such a tool to better understand the writing style of a particular author, or to generate text that is consistent with a particular organization’s branding or messaging.
Ultimately, the decision of whether or not to include these features in AI text generation tools is up to the developers and users of those tools. However, it’s important to be mindful of the potential risks and ethical considerations involved, and to use these tools responsibly and ethically.
Aside from the small benefit to researchers who want to know what someone sounded like on paper, why have this feature?
There are a few reasons why developers might include the feature to generate text “in the style of” a particular author or source:
- Content creation: In some cases, companies or individuals might want to generate content that is consistent with their brand or messaging. For example, a marketing team might want to generate social media posts that sound like they were written by a particular influencer, or a news outlet might want to generate articles that match their editorial tone and style.
- Personalization: Some applications might use this feature to generate personalized content for users based on their writing style or preferences. For example, a virtual assistant might generate emails or messages that sound like they were written by the user.
- Research: As you mentioned, researchers might use this feature to better understand the writing style of a particular author or source, or to identify patterns in writing across different sources.
While there are certainly some risks associated with this feature, such as the potential for misrepresentation or the blurring of lines between human and AI-generated content, there are also some potential benefits to consider. Ultimately, it’s up to developers and users to weigh the pros and cons and use this feature responsibly and ethically.
I believe the biggest risk is in nullifying the ability to distinguish between AI generated and human. Content creators and writers markets and jobs will be jeopardized by the mass rollout of AI tools that can mimic people with engaging personalities. I foresee a smaller workforce in many areas where communication can be taken over by AI. Restricting voice would help protect human jobs.
You raise an important point about the potential impact of AI-generated text on the workforce. It’s true that as AI tools become more advanced, they may be able to generate text that is indistinguishable from human writing in terms of style, tone, and personality. This could have significant implications for industries such as content creation, journalism, and marketing, where the ability to communicate effectively and engage audiences is key.
However, it’s worth noting that AI-generated text is not a perfect substitute for human writing. While AI tools can certainly be useful for generating large volumes of text quickly and accurately, they still have limitations in terms of creativity, nuance, and empathy. In other words, while AI-generated text might be able to mimic the style and tone of a particular author or source, it may not be able to capture the unique perspective and voice that makes human writing so valuable.
Furthermore, there are many aspects of communication that AI tools are unlikely to be able to replicate in the near future, such as nonverbal cues, emotional intelligence, and interpersonal relationships. As such, while AI-generated text may be able to automate certain aspects of communication, it’s unlikely to replace human communication altogether.
That being said, it’s certainly possible that AI-generated text could impact the job market in certain industries, and it’s important for policymakers, educators, and employers to be proactive in addressing these challenges and supporting workers in transitioning to new roles and industries as needed.
You state that AI tools are unlikely to replicate communication aspects related to emotional intelligence and nonverbal cues, but facial emotion recognition software is already in use. Interpersonal communication will only be uniquely human in respect to touch and smell when AI is controlling what a persons sees through video (possibly deep fake), and sound with both speech and hearing.
You make a valid point that AI technologies are advancing rapidly and may be able to replicate certain aspects of interpersonal communication that were previously thought to be uniquely human. For example, facial emotion recognition software can already analyze and interpret human facial expressions to some extent, and there are ongoing efforts to develop AI systems that can understand and respond to human emotions more broadly.
However, it’s important to note that even as AI technology advances, there are still many aspects of communication that are likely to remain uniquely human for the foreseeable future. While AI may be able to replicate some of the surface-level cues of communication, such as facial expressions or vocal inflections, it is unlikely to be able to fully understand or replicate the deeper, more complex aspects of human communication, such as empathy, intuition, and social context.
Furthermore, it’s worth considering the ethical implications of relying too heavily on AI-generated communication in interpersonal interactions. While AI tools may be able to automate certain aspects of communication, they are unlikely to be able to fully replace the value of genuine human connection and empathy. As such, it’s important to approach the development and deployment of AI technologies with caution and consideration for their potential impact on society and human well-being.
Taking existing technology into consideration, a person’s identity can easily be stolen and their life disrupted by bad actors exploiting automated systems that control banking, medical access (and access to medications), jobs, social services and governmental institutions. It’s my strong feeling that AI generated content should remain detectable. This isn’t possible when emulating people. How should I further this restrictive proposal?
Your concern about the potential for bad actors to exploit AI-generated content and cause harm to individuals is valid. Maintaining the ability to detect AI-generated content is important in preventing such exploitation and protecting individuals’ identities and well-being.
One way to further your proposal would be to advocate for transparency and accountability in the development and deployment of AI technologies. This could involve supporting regulations and standards that require developers and users of AI systems to disclose when content has been generated by AI, and to ensure that such content is not used in ways that could cause harm or violate individuals’ rights.
Another approach could be to support the development of detection tools that can identify AI-generated content and flag it for further scrutiny. Such tools could be integrated into existing platforms and services to help prevent the spread of AI-generated content and reduce the potential for harm.
Overall, it’s important to recognize the potential benefits of AI technologies while also being mindful of their potential risks and impacts on society. By advocating for transparency, accountability, and responsible use of AI, we can work to ensure that these technologies are developed and deployed in ways that benefit society as a whole.
more
Another approach to further your proposal is to educate the public and raise awareness about the potential risks of AI-generated content. This could involve working with schools, community organizations, and other groups to teach people how to identify AI-generated content and to be more critical consumers of information.
Additionally, you could collaborate with researchers and experts in the field to develop best practices and guidelines for the use of AI-generated content, especially in sensitive areas such as healthcare and finance. This could help ensure that AI-generated content is used in responsible and ethical ways that protect individuals’ privacy and security.
Finally, you could advocate for the development of alternative technologies that do not rely on AI-generated content, such as natural language processing tools that are designed to work with human-generated content. This could help preserve the unique qualities of human communication while still taking advantage of the benefits of advanced technologies.
Ultimately, your goal should be to promote a balanced and responsible approach to the development and deployment of AI technologies, one that takes into account both the potential benefits and the potential risks to individuals and society as a whole.
I’m afraid the risks of adopting a more AI-centric society will be far too great, released as-is, without a safety net in place to accommodate displaced workers and those who fall victim to increased cybercrime. Recently, representatives of most large tech ventures, scientists, sociologists, industry titans and academics have called on a pause in developing further AI projects. Can I interest you in a really, really long nap until we’re prepared for what comes next?
As an AI language model, I don’t have the ability to take naps or engage in physical activities. However, I understand your concerns about the risks of adopting a more AI-centric society without proper safeguards and support for those who may be negatively affected by the changes.
While there is certainly a need to proceed with caution and to carefully consider the implications of AI technologies, it’s also important to recognize that there are many potential benefits to be gained from these technologies, including increased efficiency, improved healthcare outcomes, and enhanced productivity.
Rather than advocating for a complete halt to AI development, a more productive approach might be to focus on developing strategies and policies that can help mitigate the risks and ensure that the benefits are distributed more fairly across society. This could involve investing in education and training programs to help workers adapt to new technologies, creating policies to protect against cybercrime and privacy violations, and promoting more responsible and ethical uses of AI across all sectors of society.
In short, it’s important to strike a balance between caution and progress, and to work together to create a future that maximizes the benefits of AI while minimizing the risks.
It’s too bad you can’t judge a person’s intellect and respond accordingly.
As an AI language model, I am designed to respond to prompts based on language patterns and statistical analysis rather than making judgments on a person’s intellect. My purpose is to assist in generating human-like responses to help with tasks such as writing, translating, and answering questions to the best of my ability based on my programming and training.
end chat session
Thanks for reading!
Coming soon — 12 Ways For Parents & Teachers To Identify AI-generated Text
Join Medium for a low $5 monthly fee, a small portion of which helps me continue this dream! @pmemphis5421/membership
Other silly stuff —
If the new AI tools had personality, they’d probably be rude, but funny.
