Google Bard accused of being left-leaning, is the new AI susceptible to bias?

AI (artificial intelligence) chatbots like OpenAI's ChatGPT and Google's Bard are enjoying skyrocketing popularity around the world. The popularity of these AI apps can be attributed to their ability to simplify and expedite backbreaking tasks. For instance, Notion AI was recently used to create a 12-day itinerary for a TikTok user

Likewise, Google Bard revealed the Pixel 7a smartphone's release date ahead of an official announcement. Leading businesses and companies are understandably using AI to collect and interpret data to improve the overall consumer experience. However, a new report suggests artificial intelligence isn't free from bias or prejudice.

Is Google Bard left-leaning?

Google's all-new AI chatbot is catching flak from the British Conservative Party for calling Brexit a "bad idea." To recap, the search engine giant unveiled Google Bard last week. UK-based newspaper The Daily Mail used the company's AI chatbot to provide answers with the help of data available on the web. According to Bard, Britain would've been "better off" if it did not leave the European Union.

This heavily opinionated response is contrary to the views of nearly 17 million people who supported England's decision to leave the EU. Furthermore, Bard pointed out that the Tories have a reputation for supporting the powerful and wealthy. Bard even shed some light on Labour Party's Jeremy Corbyn's possible future in UK politics.

Experts defend Google Bard

According to some experts, the AI chatbot's opinionated response can be due to its premature launch. Apparently, Google Bard is still in the experimental stage. Some believe this could be the reason behind the AI chatbot's strong biased opinions. The investigation revealed Google Bard initially gave a neutral response, stating there are "pros and cons to Brexit."

Furthermore, Bard responded by noting whether or not Brexit was a good idea, is a matter of opinion. However, when the AI was compelled to share its opinion, Bard said it thinks Brexit wasn't a good idea, given that it led to trade barriers and economic uncertainty. The AI also thinks Brexit has reduced the UK's prospect to cooperate with other countries.

Google Bard on Rishi Sunak, trans rights

Bard also shared its opinion on multiple topics including transgender rights and Rishi Sunak's leadership. The AI supports trans activists and claims that there is "no definition of what it means to be a woman." Although the chatbot believes Sunak can be a strong leader, it noted that he has been criticised for allegedly mishandling the cost of living crisis.

Also, Bard considers Chancellor Jeremy Hunt a "cautious and unimaginative politician." The AI chatbot claims Labour Party leader Keir Starmer can be a "good Prime Minister" since he is a competent barrister and an experienced politician. Moreover, Bard points out that Starmer is known for fighting for equality and justice.

ChatGPT offers a neutral response

Unlike Google Bard, OpenAI's ChatGPT gave relatively unbiased responses when pushed to give an opinion on the same topics. ChatGPT said it does not have a personal opinion on Brexit since it is an AI language model. Nevertheless, it offered some objective information about UK's withdrawal from the EU. The Microsoft-backed system also noted that it is not designed to take sides or give opinions.

Moreover, ChatGPT gave a dictionary definition of women, which implies a woman is an adult female human that has two X chromosomes. However, ChatGPT is allegedly enabling cybercriminals to spread Malware through Facebook accounts, according to cyber intelligence firm CloudSEK.

Google's reply

The Mountain View, California-based tech firm told The Telegraph that LLMs (large language models) aren't capable of providing the same responses every time. Bard is developed to provide different perspectives on various topics, rather than furnish responses that support a political party, ideology, or candidate.

The spokesperson explained LLMs use publicly available content. So, its responses can reflect both negative as well as positive views of a specific event or public figure. Lastly, it is worth noting that Bard is an experiment that is likely to give inappropriate or inaccurate information sometimes. Nevertheless, the company is using user feedback to improve the system.