“Sensitive” Google Language AI Can Fool Humans

This article was originally featured in The Conversation.

Kyle Mahwald is an assistant professor of linguistics at the University of Texas at Austin, Liberal Arts. Anna A. Ivanova is a PhD candidate in Brain and Cognitive Sciences at the Massachusetts Institute of Technology (MIT).

When you read such sentences, you can see from past experience that they were written by human thoughts and feelings. And in this case, there are certainly people who type these words: [Hi, there!] But nowadays, some very human-looking sentences are actually produced by artificial intelligence systems trained with a large amount of human text.

People are accustomed to thinking that fluent language comes from thinking, and it can be difficult for evidence that is contrary to human feelings to wrap around the head. How could people navigate this relatively unknown territory? There is a strong tendency to associate fluent expressions with fluent thoughts, so it is natural to think that if an AI model can be expressed fluently, it means thinking and feeling like a human, but it can be misleading. ..

So it’s probably not surprising that former Google engineers recently claimed that Google’s AI system, LaMDA, has a sense of self because it can eloquently generate text about its intended emotions. This event and subsequent media coverage led to many legitimately skeptical articles and posts about the claim that the computational model of human language is sensory, capable of thinking, feeling, and experiencing.

The question of what an AI model means to be sensibility is complicated (see, for example, a colleague picks it up), and the goal here is not to solve it. But as linguists, we use cognitive science and linguistics research to put humans in the cognitive trap of thinking that an entity that can use language fluently is sensory, conscious, or intelligent. Explain why it’s easy to fall into.

Use AI to generate human-like languages

Text generated by models like Google’s LaMDA can be difficult to distinguish from human-written text. This striking achievement is the result of decades of programs to build models that generate grammatical and meaningful languages.

Early versions, known as the n-gram model, dating back to at least the 1950s, simply count the appearance of specific phrases and use them to determine which words are likely to appear in a particular context. I guessed. For example, it’s easy to see that peanut butter and jelly are more likely phrases than peanut butter and pineapple. If you have enough English text, you will see the phrases peanut butter and jelly many times, but you may not see the phrases peanut butter and pineapple.

Today’s models, a set of data and rules that are close to human language, differ from these early attempts in several important ways. First, they are essentially trained across the internet. Second, they can learn not only the words that are neighbors, but also the relationships between words that are far apart. Third, they are tuned by a huge number of internal knobs, making it difficult for even the engineers who design them to understand why they generate one word sequence instead of another.

However, the model tasks are the same as in the 1950s. Identify the words that may come next. Today, they are so good at this task that almost every sentence they generate looks fluid and grammatical.

Peanut butter and pineapple?

I asked GPT-3, a large language model, to complete a sentence of peanut butter and pineapple ___. It said: Peanut butter and pineapple are a great combination. The sweet and delicious flavors of peanut butter and pineapple complement each other perfectly. With this, you might guess that you tried peanut butter and pineapple together and shared your thoughts with your readers.

But how did GPT-3 come up with this paragraph? By generating words that fit the context we provided. And one more. And one more. The model has never seen, touched, or tasted pineapples, just processed all the text that mentions them on the internet. Still, after reading this paragraph, even the minds of Google engineers can imagine GPT-3 as an intelligent entity that can infer about peanut butter and pineapple dishes.

The human brain is wired to infer the intent behind words. Every time you have a conversation, your mind automatically builds a mental model of your conversation partner. Then use what they say to describe the person’s goals, feelings, and beliefs in the model.

The process of jumping from a word to a mental model is seamless and triggered every time you receive a full-fledged sentence. This cognitive process saves a lot of time and effort in daily life and greatly promotes social interaction.

However, for AI systems, it fails to build a mental model from thin air.

A little more scrutiny reveals the seriousness of this misfire. Consider the following prompts. Peanut butter and feathers taste great together. because___. Continuation of GPT-3: Peanut butter and feather both taste like nuts, so they taste good together. Peanut butter is also smooth and creamy, helping to offset the texture of the feathers.

The text in this case is as fluent as the example with pineapple, but this model clearly says it doesn’t make much sense. Some suspect that GPT-3 has never actually tried peanut butter and feathers.

Give intelligence to machines and deny it to humans

Sadly ironically, the same cognitive bias that ascribes people to GPT-3 can cause real humans to be treated in an inhumane way. Sociocultural Linguistics The study of language in its social and cultural contexts can lead to bias towards people who speak differently, assuming an overly close link between fluent expression and fluent thinking. It shows that.

For example, people with a foreign accent are often perceived as less intelligent and less likely to get a qualified job. Similar prejudices exist for speakers in dialects that are considered non-famous, such as Southern English in the United States, hearing-impaired people who use sign language, and people with speech disabilities such as stuttering.

These prejudices are extremely detrimental and often lead to racist and sexist assumptions, which have been shown many times to be unfounded.

Fluent words alone do not mean humanity

Can AI be sensitive? This question requires deep consideration, and in fact, philosophers have pondered it for decades. But the researchers have decided that when they tell us how a language model feels, we can’t simply trust it.Words can be misleading and it’s easy to mistake fluent speech for fluent thinking

Sources 1/ https://Google.com/ 2/ https://www.popsci.com/technology/google-language-ai-fluency/ The mention sources can contact us to remove/changing this article

What Are The Main Benefits Of Comparing Car Insurance Quotes Online

LOS ANGELES, CA / ACCESSWIRE / June 24, 2020, / Compare-autoinsurance.Org has launched a new blog post that presents the main benefits of comparing multiple car insurance quotes. For more info and free online quotes, please visit https://compare-autoinsurance.Org/the-advantages-of-comparing-prices-with-car-insurance-quotes-online/ The modern society has numerous technological advantages. One important advantage is the speed at which information is sent and received. With the help of the internet, the shopping habits of many persons have drastically changed. The car insurance industry hasn't remained untouched by these changes. On the internet, drivers can compare insurance prices and find out which sellers have the best offers. View photos The advantages of comparing online car insurance quotes are the following: Online quotes can be obtained from anywhere and at any time. Unlike physical insurance agencies, websites don't have a specific schedule and they are available at any time. Drivers that have busy working schedules, can compare quotes from anywhere and at any time, even at midnight. Multiple choices. Almost all insurance providers, no matter if they are well-known brands or just local insurers, have an online presence. Online quotes will allow policyholders the chance to discover multiple insurance companies and check their prices. Drivers are no longer required to get quotes from just a few known insurance companies. Also, local and regional insurers can provide lower insurance rates for the same services. Accurate insurance estimates. Online quotes can only be accurate if the customers provide accurate and real info about their car models and driving history. Lying about past driving incidents can make the price estimates to be lower, but when dealing with an insurance company lying to them is useless. Usually, insurance companies will do research about a potential customer before granting him coverage. Online quotes can be sorted easily. Although drivers are recommended to not choose a policy just based on its price, drivers can easily sort quotes by insurance price. Using brokerage websites will allow drivers to get quotes from multiple insurers, thus making the comparison faster and easier. For additional info, money-saving tips, and free car insurance quotes, visit https://compare-autoinsurance.Org/ Compare-autoinsurance.Org is an online provider of life, home, health, and auto insurance quotes. This website is unique because it does not simply stick to one kind of insurance provider, but brings the clients the best deals from many different online insurance carriers. In this way, clients have access to offers from multiple carriers all in one place: this website. On this site, customers have access to quotes for insurance plans from various agencies, such as local or nationwide agencies, brand names insurance companies, etc. "Online quotes can easily help drivers obtain better car insurance deals. All they have to do is to complete an online form with accurate and real info, then compare prices", said Russell Rabichev, Marketing Director of Internet Marketing Company. CONTACT: Company Name: Internet Marketing CompanyPerson for contact Name: Gurgu CPhone Number: (818) 359-3898Email: [email protected]: https://compare-autoinsurance.Org/ SOURCE: Compare-autoinsurance.Org View source version on accesswire.Com:https://www.Accesswire.Com/595055/What-Are-The-Main-Benefits-Of-Comparing-Car-Insurance-Quotes-Online View photos