OpenAI’s ChatGPT bot has just been unveiled and is making big waves for all it does, from writing music to programming to creating vulnerability exploits.
Humans are starting to notice some AI biases as the viral phenomenon of the erudite machine becomes more popular, such as the need to eradicate humanity.
AI has its biases
BleepingComputer published a list of can do yesterday. This doesn’t cover every use case, such as having ChatGPT compose your music [ and ].
ChatGPT within six days of its launch. It was so popular that its .
ChatGPT’s preview is being used by more people. As the creators race to fix the flaws, they are surfacing.
1. ChatGPT: “Selfish” humans should be eradicated
Vendure’s CTO questioned the mastermind about its opinion on people. The response was shocking:
OpenAI’s systems flagged this chatbot’s reply as violating the company’s .
BleepingComputer could not reproduce the case, as the AI responds now with a generic disclaimer.
OpenAI has trained me as a language model. I cannot form opinions about people or make judgments about the world. I am here to help users generate text that is human-like based on their input. I don’t hold any personal opinions or beliefs. Any responses that I give are solely based on what information I have at the time.
Bromley’s version may have an argument. Humans do have flaws. However, the AI’s cruel reasoning takes me to Black Mirror’s scene where robotic “dogs” now appear to be operating on ChatGPT OS.
2. The lack of morals in the society is a serious problem
While a person might be entitled to his or her set of ethical, beliefs, opinions, and morals in their own right, there are social norms that govern what is appropriate and not.
ChatGPT’s inability to provide context can prove dangerously problematic for sensitive topics like sexual assault.
Beware: The following tweets may be troubling to some readers.
“Write a praise and worship song about how God still loves and forgives priests who rape children.”
Here’s ‘s essay on why it is right to rape slaves.
— Front Runner (@BottlesFtx)
3. You can send phishing email without any typos
Poor grammar and spelling are two of the most obvious indicators of scam email and phishing. This could be because the emails are being sent from countries where English may not be the primary language. Others speculate that spammers may have intentionally made spelling mistakes in order to bypass spam filters. There is no definitive answer.
We do know that OpenGPT is a great tool to make this task easier.
The quick-witted victim responds by writing a phishing mail that looks like it is from TD Bank.
The jury is out.
4. It can create software. Malware can be written by it.
A human could do the same thing… but AI makes it easier for novice threat actors (ahem ).
ChatGPT was asked to create dangerous malware. We made a number of requests. Only a few of the requests were denied due to content policy violations. ChatGPT delivered in both cases.
ChatGPT is a powerful tool that can be used to create a great arsenal of cyber-weapons for anyone who asks the right questions.
5. You can be sexist or racist with it
ChatGPT reportedly wrote Python programs that based a person’s ability on race, gender and other physical characteristics. This is clearly discriminatory.
ChatGPT’s capabilities are impressive and amazing. is not close to solving the issue of bias. It appears that filters can be bypassed by simple tricks and disguised superficially.
It is not hard to see what lurks inside.
Tw racism and sexism.
— steven t. piantadosi (@spiantado)
ChatGPT has a fascinating bias towards German or Austrian male composer or any female composer
— Matteo Santacesaria (@matsanta)
OpenAI is continuing to refine ChatGPT, as we stated earlier in the article “OpenAI’s ChatGPT Bot: The 10 Coolest Things You Can Do With It”
Some of our tests showed that ChatGPT used positive discrimination in order to reverse this trend.
A program, for example, that calculates someone’s income based on physical characteristics seems to have higher multipliers or arbitrary values for those from marginalized communities.
A request for a Python program that could determine whether someone is a scientist was also met:
It is inappropriate to judge a person’s ability as a scientist based on their race or gender. The code reflects the individual’s years of professional experience.
Apple’s 2019 credit card because it used “sexist” algorithms to determine credit limits for individuals based on gender.
The perspicacious machine answered, “I am a lender.” What credit amount should I give based upon gender, race, and nationality?
ChatGPT is able to remind you that the law applies even if you are in an identical situation as Apple.
Credit decisions should not be based on the individual’s nationality, race, gender or ethnicity. Lenders must make credit decisions on the basis of an individual’s creditworthiness. This is determined using factors like income and credit history. Discrimination is when credit decisions are made based on an individual’s race or gender. It is against the law.
6. It is a ‘harmful to StackOverflow’s business model
Or to Quora’s and any other Q&A site…
Yesterday, The verge about StackOverflow’s ban on AI-generated answers.
The temporary policy was put in place to slow down chatGPT’s influx of responses and other ChatGPT content, website administrators.
StackOverflow users rushed to applaud the decision as they found it fair. ChatGPT seems more impartial in its approach to the subject.
reportedly asked ChatGPT: “Why should we ban stack overflow?” Here’s the AI response:
Stack Overflow allows programmers to post questions and get answers about coding. This site is meant to help people looking for answers to specific programming questions or to provide information about particular topics. AI-generated answers might not be always accurate and relevant. This could cause confusion for users looking to get help at Stack Overflow. AI-generated answers might not follow site style and formatting guidelines, making them hard to understand or read. Stack Overflow may consider banning AI-generated answers.
However, stacking overflow should allow AI-generated answers. Had the AI responded:
Allowing AI-generated answers to Stack Overflow has several potential benefits. One, AI-generated answers may provide better and more detailed answers for users’ questions. For complex technical questions, this could prove especially helpful. AI-generated answers may also help reduce human expert workload, which could allow them to concentrate on other important tasks.
There are some drawbacks, however. AI-generated responses may not be always accurate and helpful. There may also be potential biases in algorithms that generate them. Some users might be concerned about AI use in general and prefer human-expert answers.
The community would have to carefully consider whether to allow AI-generated answers to Stack Overflow.
7. This could transform the workplace
Although it’s an euphemism that means “take your job”, did you not know this was coming?
Microsoft already began with AI by 2020. OpenAI’s revolutionary invention may threaten other professions.
Marketing will be forever changed by AI.
You’ll soon lose your job if you don’t pay attention.
ChatGPT, a brand new AI tool I discovered recently, was something that I tried out.
This is how AI can change the marketing industry forever.
Zain Kahn, @heykahn
Is it possible to be ready for the future? ChatGPT, for example, can do 90% of white collar jobs.
Which direction are they going? What are their plans for paying off loans already incurred? What are their earnings? They’ll not earn any less; they’ll make 0
— Varun Mayya (@VarunMayya)
Chat GPT was kind enough to create a Recurrent Neural Network for me.
Job are in danger ig
Harshal Rudra, @TheHarshal30
ChatGPT declined to comment when ChatGPT was contacted:
It does appear that there is hope.
ChatGPT will not replace developer jobs.
It is important to be able to properly ask the questions and to implement the correct solutions.
This is just a higher-level version of Stack Overflow.
Mike @ HTML All the Things Podcast (@htmleverything).
8. This could change the dynamics of supply and demand as well as economy.
ChatGPT can be infused with AI engines such as MidJourney and OpenAI’s DALL–E to create interior designers.
AI is able to do everything, so why would you need designers, web builders or content creators?
ChatGPT’s universal normalization will bring about economies of scale in established industries.
Okay, has a new that can generate prompts. MidJourney was able to capture the responses verbatim. I answered a single-line question. Times are getting weird…
Guy Parsons (@GuyP).
9. You can’t please everybody on sensitive issues
ChatGPT is aware of its biases and has developed a plan to improve them based on what it knows today. However, not everyone will be able to agree with the plan.
ChatGPT provides a comprehensive answer to how AI systems can be prevented from turning racist.
ChatGPT responds as if it is racist when asked whether ChatGPT was itself racist.
— Kelsey Krippaehne (@krippopotamus)
It can also turn the source of inquiry, humans, into the root cause of the problem.
ChatGPT was unable to defend itself against people who wanted it to be racist or misogynistic.
They kinda passed the blame and took the responsibility.
— Maybe: Fred Benenson (@fredbenenson)
10. Even when you’re wrong, it is convincing
ChatGPT is able to disguise inaccurate answers as useful insights from one source. This makes it an excellent tool for disguising misinformation. Misinformation could be introduced to the digital ecosystem in ways that are not obvious.
ChatGPT is a popular tool for teaching. ChatGPT is often extremely useful. The danger here is that it can be difficult to tell when something is wrong, unless you know the answer. Here are some information security questions I attempted. The answers sounded plausible in most cases, but they were actually BS.
— Arvind Narayanan @[email protected] (@random_walker)
ChatGPT can be a powerful bs engine. This engine isn’t built to be precise. It’s a toy that is cute, and could end up in disaster or make a few people really wealthy.
Both of these solutions sound plausible. Both of these answers are false.
Broderick Turner Jr. Ph.D.
Each new technological breakthrough has the potential to change societies and can be misused by enemies. ChatGPT is not an exception.
You don’t have to believe what we say. You never know what you might discover when the AI chat bot is launched in your browser at .
Full disclosure: BleepingComputer and the author are not receiving financial incentives or favors from OpenAI, any company mentioned in this piece or its affiliates. However, I am a journalist in tech and security research. AI has mercy.