Serendipity35

I read an article from Rutgers University-Camden that begins by saying that most people now realize that artificial intelligence has become increasingly embedded in everyday life. This has created concerns. One of those lesser-spoken-about concerns is around bias in its programming. Here are some excerpts from the article.

Some AI tasks are so innocuous that users don't think about AI being involved. Your email uses it. Your online searches use it. That smart thermostat uses it. Are those uses frightening? But as its capabilities expand, so does its potential for wide-ranging impact.

Organizations and corporations have jumped on the opportunity presented by AI and Big Data to automate processes and increase the speed and accuracy of decisions large and small. Market research has found that 84 percent of C-suite executives believe they must leverage artificial intelligence to achieve their growth objectives. Three out of four believe that if they don't take advantage of AI in the next five years, they risk going out of business entirely.

Bias can cause artificial intelligence to make decisions that are systematically unfair to particular groups of people, and researchers have found this can cause real harm. The Rutgers–Camden researcher Iman Dehzangi, says that “Artificial intelligence and machine learning are poised to create valuable opportunities by automating or accelerating many different tasks and processes. One of the challenges, however, is to overcome potential pitfalls such as bias.” 

What does biased AI do? It can give consistently different outputs for certain groups compared to others. It can discriminate based on race, gender, biological sex, nationality, social class, or many other factors.

Of course it is human beings who choose the data that algorithms use and humans have biases whether they are conscious of them or not.

"Because machine learning is dependent upon data, if the data is biased or flawed, the patterns discerned by the program and the decisions made as a result will be biased, too," said Dehzangi, pointing to a common saying in the tech industry: "garbage in, garbage out." “There is not a successful business in operation today that is not using AI and machine learning,” said Dehzangi. Whether it is making financial investments in the stock market, facilitating the product development life cycle, or maximizing inventory management, forward-thinking businesses are leveraging this new technology to remain competitive and ahead of the curve. However, if they fail to account for bias in these emerging technologies, they could fall even further behind, remaining mired in the flawed data of the past. Research has revealed that if care is not taken in the design and implementation of AI systems, longstanding social biases can be embedded in the systems' logic.

Tay was a chatbot originally released by Microsoft Corporation as a Twitter bot on March 23, 2016. It "has had a great influence on how Microsoft is approaching AI," according to Satya Nadella, the CEO of Microsoft.

Tay caused almost immediate controversy when the bot began to post inflammatory and offensive tweets through its Twitter account, causing Microsoft to shut down the service only 16 hours after its launch. According to Microsoft, this was caused by trolls who "attacked" the service as the bot made replies based on its interactions with people on Twitter - a dangerous proposition.

It was named "Tay" as an acronym for "thinking about you." It was said that it was similar to or based on Xiaoice, a similar Microsoft project in China that Ars Technica reported that it had "more than 40 million conversations apparently without major incident".

Interestingly, Tay was designed to mimic the language patterns of a 19-year-old American girl and presented as "The AI with zero chill."

It was quickly abused with Twitter users began tweeting politically incorrect phrases, teaching it inflammatory messages so that the bot began releasing racist and sexually-charged messages in response to other Twitter users.

One artificial intelligence researcher, Roman Yampolskiy, commented that Tay's misbehavior was understandable because it mimicked the deliberately offensive behavior of other Twitter users, and Microsoft had not given the bot an understanding of inappropriate behavior. He compared the issue to IBM's Watson, which began to use profanity after reading entries from the website Urban Dictionary.

It was popular in its short life. Within 16 hours of its release, Tay had tweeted more than 96,000 times, That is when Microsoft suspended the account for "adjustments." Microsoft confirmed that Tay had been taken offline, released an apology on its official blog, and said it would "look to bring Tay back only when we are confident we can better anticipate malicious intent that conflicts with our principles and values."

Then on March 30, 2016, Microsoft accidentally re-released the bot on Twitter while testing it. Given its freedom, Tay released some drug-related tweets, then it became stuck in a repetitive loop of tweeting "You are too fast, please take a rest", several times a second. The posts appeared in the feeds of 200,000+ Twitter followers.

Tay has become a cautionary tale on the responsibilities of creators for their AI.

In December 2016, Microsoft released Tay's successor, a chatbot named Zo which was an English version of Microsoft's other successful chatbots Xiaoice (China) and Rinna [ja] (Japan).

Terms of service. That information you tend to avoid reading. Good example: Google's newly updated terms of service, which I found out about in an email last week. I decided to read them.

Their updated terms opens with "We know it’s tempting to skip these Terms of Service, but it’s important to establish what you can expect from us as you use Google services, and what we expect from you. These Terms of Service reflect the way Google’s business works, the laws that apply to our company, and certain things we’ve always believed to be true. As a result, these Terms of Service help define Google’s relationship with you as you interact with our services."

Here are a few items I noted:
Some things considered to be abuse on the part of users includes accessing or using Google services or content in fraudulent or deceptive ways, such as:
phishing
creating fake accounts or content, including fake reviews
misleading others into thinking that generative AI content was created by a human
providing services that appear to originate from you (or someone else) when they actually originate from us
providing services that appear to originate from us when they do not
using our services (including the content they provide) to violate anyone’s legal rights, such as intellectual property or privacy rights
reverse engineering our services or underlying technology, such as our machine learning models, to extract trade secrets or other proprietary information, except as allowed by applicable law
using automated means to access content from any of our services in violation of the machine-readable instructions on our web pages (for example, robots.txt files that disallow crawling, training, or other activities)
hiding or misrepresenting who you are in order to violate these terms
providing services that encourage others to violate these terms

Take that second item I highlighted about misleading others into thinking that generative AI content was created by a human, Does that mean that if I use their generative AI or some other provider's AI to help write a blog post that I put here with my name that I am violating their terms of service?

Though I would say that Google's Terms of Service is written in plain langauage that most readers should be able to understand, the implications of some of the terms are much harder to interpret.

NOTE: The Google Terms of Service (United States version) that I reference are effective May 22, 2024.
View Archived versions and  Download a PDF

 

 

Bias has always existed. It has always existed online. Now, with AI, there is another level of bias.

Bias generated by technology is “more than a glitch,” says one expert.

For example, why does AI have a bias against dark skin? It is because its data is scraped from the Internet, and the Internet is full of biased content.

This doesn't give AI a pass on bias. It is more of a comment or reflection on bias in general.