A new report shows that AI bot DeepSeek has an 83% fail rate for accurate responses to news and information prompts.

Reuters

Karah Rucker Anchor/Reporter

Tech

Deepseek chatbot has 83% ‘fail rate’ in accuracy: NewsGuard study

Jan 29

Listen

Karah Rucker Anchor/Reporter

By Karah Rucker (Anchor/Reporter), Ian Kennedy (Lead Video Editor)

The Chinese artificial intelligence company DeepSeek has been generating a lot of buzz in the tech world, especially with its newly released chatbot. Some investors wonder if it’s a stronger competitor than Microsoft’s ChatGPT. As a result, tech stocks have experienced significant volatility this week.

DeepSeek’s rise in popularity

Despite DeepSeek’s rapid rise, with its chatbot becoming the most downloaded app on Apple’s App Store shortly after its rollout, a new study suggests that the chatbot lags behind competitors in accuracy.

DeepSeek claims its AI model performs on par or better than Western competitors like ChatGPT, owned by Microsoft-backed OpenAI, at a fraction of the cost.

Download the SAN app today to stay up-to-date with Unbiased. Straight Facts™.

Point phone camera here

The company reports that training its model cost $6 million, compared to the billions spent by Silicon Valley companies on their models.

The reported success of DeepSeek has had a major impact on U.S. tech stocks, wiping out approximately 1 trillion dollars in value.

Study highlights DeepSeek’s accuracy issues

A study conducted by NewsGuard, a global leader in information reliability, revealed issues with DeepSeek’s answers.

The research shows that DeepSeek’s chatbot only provided accurate responses to news and information prompts 17% of the time. In 30% of the cases, it repeated false claims, and 52% of the time, it gave “vague or unhelpful responses” to news-related questions.

NewsGuard rates DeepSeek’s chatbot with a fail rate of 83%, compared to 62% for its competitors.

Concerns over lack of safeguards

Further concerns about DeepSeek’s chatbot have emerged from the cybercrime threat intelligence platform KELA. The platform warned that the chatbot lacks certain safeguards and could potentially instruct users on how to make a bomb.

Tech analysts focus on cost over accuracy

Despite these issues, some analysts argue that the chatbot’s appeal lies not in its accuracy but in its low cost.

A spokesperson from D.A. Davidson told Reuters, “The importance of the DeepSeek breakthrough is not in answering news-related questions accurately; it is in the fact that it can answer any question at 1/30th of the cost of comparable AI models.”

NewsGuard rating of Straight Arrow News

NewsGuard, the organization behind the study, is a leader in providing data and analysis on the reliability of news sources. It helps guide consumers toward trustworthy journalism by evaluating news organizations based on nine factors of credibility.

NewsGuard has given Straight Arrow News a 100% reliability rating for its unbiased and fact-based reporting.

Click here to learn more about SAN’s mission and early successes in reaching trustworthy standards from media watchdog groups.

Tags: Artificial Intelligence, Big Tech, ChatGPT, China, cyber crimes, Cybersecurity, DeepSeek, Microsoft, Tech, Technology

THE CHINESE RIVAL AI COMPANY “DEEPSEEK” IS ALL THE BUZZ.

IT’S MADE SOME INVESTORS WONDER IF THE NEW “DEEPSEEK CHATBOT” IS BETTER THAN MICROSOFT’S CHAT GPT –

SENDING TECH STOCKS ON A ROLLERCOASTER THIS WEEK.

BUT ONE STUDY IS SUGGESTING – IT LAGS FAR BEHIND ITS COMPETITORS – IN ACCURACY.

DEEPSEEK’S CHATBOT BECAME THE MOST DOWNLOADED APP IN APPLE’S APP STORE WITHIN DAYS OF ITS ROLL-OUT.

THE QUICK POPULARITY – COMES AS DEEPSEEK CLAIMED IT PERFORMS “ON PAR OR BETTER” THAT WESTERN COMPETITORS – SUCH AS CHATGPT – OWNED BY MICROSOFT-BACKED OPEN A-I –

AT A **FRACTION OF THE COST.

DEEP SEEK’S MODEL COST 6 MILLION TO TRAIN ACCORDING TO THE COMPANY.

COMPARED TO BILLIONS THAT SILICON VALLEY COMPANIES ARE SPENDING ON THEIR MODELS.

DEEP SEEK’S REPORTED SUCCESSES – LED TO ABOUT ONE TRILLION DOLLARS BEING WIPED IN U.S. TECH STOCKS.

WHILE IT’S BEEN A LOT OF GOOD NEWS FOR DEEPSEEK –

A NEW STUDY CONDUCTED BY “NEWS GUARD” SUGGESTS –

ITS AI CHATBOT ONLY ACHIEVES “ACCURACY” IN ITS RESPONSES TO NEWS AND INFORMATION PROMPTS **17 PERCENT OF THE TIME.

30% OF THE TIME – IT REPEATED FALSE CLAIMS.

AND 52% OF THE TIME IT GAVE VAGUE OR NOT USEFUL RESPONSES TO NEWS RELATED QUESTIONS.

NEWS GUARD – GIVING DEEPSEEK AN “83%” FAIL RATE.

THAT’S COMPARED TO A 62% FAIL RATE GIVEN TO AI CHATBOT RIVALS –

AFTER NEWS GUARD FED THE BOTS THE SAME 300 PROMPTS – AND DOCUMENTED RESPONSES.

A CYBERCRIME THREAT INTELLIGENCE PLATFORM CALLED “KELA” ALSO WARNED DEEPSEEK LACKS CERTAIN SAFEGUARDS – SAYING THE CHATBOT COULD INSTRUCT A USER ON HOW TO MAKE A BOMB.

BUT DESPITE SOME DAMNING REPORTS –

SOME TECH ANALYSTS SAY –

BEING ACCURATE OR NOT – ISN’T WHAT MAKES DEEP SEEK SO APPEALING.

WHAT MATTERS – IS THE MONEY.

A “D.A. DAVIDSON SPOKESPERSON” TELLING REUTERS –

“The importance of the DeepSeek breakthrough is not in answering news-related questions accurately, it is in the fact that it can answer any question at 1/30th of the cost of comparable AI models.”

NEWS GUARD – THE GROUP THAT CONDUCTED THE STUDY –

CALLS ITSELF A “GLOBAL LEADER IN INFORMATION RELIABILITY.”

IT PROVIDES DATA AND ANALYSIS – LIKE THE STUDY AT THE FOREFRONT OF THIS STORY.

IT ALSO HELPS GUIDE AMERICANS TOWARD TRUSTWORTHY AND RELIABLE JOURNALISM.

NEWS GUARD RATES NEWS ORGANIZATIONS BASED ON NINE FACTORS INVOLVING CREDIBILITY.

THE MEDIA WATCHDOG HAS RATED STRAIGHT ARROW NEWS AS HAVING A 100% “NEWS RELIABILITY RATING.”

UNBAISED, STRAIGHT FACT REPORTING – IS SOMETHING WE’RE PROUD OF –

AND SOMETHING WE STRIVE TO MAINTAIN AS WE PRESENT THE STORIES THAT MATTER TO YOU IN A FAIR AND FACTUAL WAY.

YOU CAN LEARN MORE ABOUT OUR NEWS ORGANIZATION BY DOWNLOADING THE STRAIGHT ARROW NEWS MOBILE APP.

OR VISIT US ONLINE AT SAN DOT COM.