AI’s Growing Influence on Workplaces Sparks Concerns Over Declining Critical Thinking

Researchers from Microsoft Research and Carnegie Mellon University studied knowledge workers by analyzing AI tools with 1,000 real-world examples to determine whether AI is changing our critical thinking.

The results of the study showed that AI is changing how we think at work and impacting our job satisfaction and related challenges. The researchers studied how knowledge workers apply critical thinking while working with AI and what the causes are that make them think more or less while using AI tools.

The study found that the more people trust AI, the less they question its results, and those who think that their skills are better than AI tend to think critically about AI responses. This is also creating a risk because as AI is getting better, we are becoming less likely to question its output even when a little brainpower needs to be flexed. There are different factors that are preventing people from thinking critically, like awareness barriers, motivation barriers, and ability barriers.

Senior researcher at Microsoft, Lev Tankelevitch, says that most people are less critical about AI output when tasks they perform using AI are low stakes, and they naturally become critical when tasks are high-stakes. The main concern is that if workers are not using critical thinking regularly while using AI, they may forget to use it even when it truly matters. There is no doubt that generative AI has made cognitive tasks easier, like helping workers in areas like comprehension, knowledge gathering, and analysis.

Even though AI is retrieving information quickly, professionals should focus on checking accuracy. AI is also being used for problem-solving, but it is important for workers to refine and adapt to solutions in real-world scenarios. Professionals are also supervising AI to ensure that it gives high-quality and relevant results instead of performing tasks on their own. As AI roles are evolving in workplaces, jobs are going to shift towards AI prompt engineering and quality control as well as output verification. The level of success in workplaces will be defined by how employees direct and access AI instead of just personal task execution. To make sure workers are using critical thinking while using AI, organizations should integrate verification steps into workflows and design AI interfaces that force users to critically analyze every response. The skills needed in AI-driven workplaces are also evolving but critical thinking skills remain important.

Image: DIW-Aigen

Read next: Social Media Abstinence Lowers Anxiety, But Mindful Usage Helps Prevent Loneliness, Researchers Discover
by Arooj Ahmed via Digital Information World

Can You Trust AI for Medical Advice? New Study Uncovers the Risky Truth

According to a new study published in NPJ Digital Medicine, some Spanish researchers tried to investigate if the large language models are reliable when it comes to giving health advice. The researchers tested seven LLMs, including OpenAI's ChatGPT, ChatGPT-4 and Meta's Llama 3, with 150 medical questions, and the researchers found that all the models tested had varied results. Most of the AI-based search engines give incomplete or incorrect results when users ask them some health-related questions. Even though AI-powered chatbots are increasingly in demand there haven't been proper studies which could show that LLMs give reliable medical-related results. This study found that the results of LLMs accuracy depend on the phrasing, retrieval bias, and reasoning, but they can still produce misinformation.

For the study, the researchers assessed four search engines: Google, Yahoo!, DuckDuckGo and Bing, and seven LLMs including ChatGPT, GPT-4, Flan-T5, Llama3 and MedLlama3. The results showed that ChatGPT, GPT-4, Llama3 and MedLlama3 had the upper hand in most evaluations, while Flan-T5 lagged behind the pack. For search engines, the researchers analyzed the top 20 ranked results. A passage extraction model was used to identify relevant snippets and a reading comprehension model was used to determine if the snippets had a definitive yes/no answer. Two types of users' behaviors were also seen: Lazy users stopped searching as soon as they found the first clear answer, while the diligent users cross-referenced three sources before deciding on an answer. The lazy users were the ones who got the most accurate answers, which shows that top-ranked answers are accurate most of the time.

• Also read: Social Media Abstinence Lowers Anxiety, But Mindful Usage Helps Prevent Loneliness, Researchers Discover

For large language models, the researchers used different prompting strategies like asking a question without any context, using friendly wording, and using expert wording. The study also provided LLMs some sample Q&As which helped some models but didn't have any effect on others. Retrieval-augmented generation method was also used where LLMs were provided search engine results before they generated their own responses. The performance of the AI models was measured through accuracy, common errors in their responses, and improvements through retrieval augmentation.

The results of the study showed that search engines answered 50-70% queries accurately while LLMs had an 80% accuracy rate. The responses from LLMs varied on the basis of how questions were framed, and the expert prompt (using expert tone) was the most effective but sometimes resulted in less definitive answers. Bing had the most reliable answers, but it wasn't any better than Yahoo!, Google, and DuckDuckGo. Many search results from search engines were irrelevant or off-topic while the precision improved 80-90% by filtering for relevant answers. Smaller LLMs showed improvements in their performance after search engine snippets were added. But poor quality retrieval worsened the accuracy of LLMs, especially for Covid-19 related queries.

The error analysis of LLMs showed that there were three major failures of LLMs when it comes to health-related queries: Incorrect medical consensus understanding, misinterpreting questions, and ambiguous answers. The study showed that the performance of LLMs varied based on the dataset they were being questioned from, with a dataset from 2020 generating more accurate responses than a dataset from 2021.

Read next: AI Search Traffic Jumps 123% as ChatGPT and Perplexity Reshape SMB Strategies
by Arooj Ahmed via Digital Information World

Social Media Abstinence Lowers Anxiety, But Mindful Usage Helps Prevent Loneliness, Researchers Discover

According to a new study from the University of British Columbia, we can still protect our mental health without quitting social media completely. We can do this by prioritizing meaningful connections over mindless scrolling on social media. Doing a complete digital detox doesn't seem realistic and possible for many people because of modern work life, so this study proves that you can still keep your mental health healthy while using social media by stopping yourself from mindless scrolling.

Many young adults use social media, which gives them some advantages as well as disadvantages, as social media is helping people to stay connected to different communities. But prolonged use of social media can also lead to increased anxiety, depression, and other mental health issues, which can affect the daily lives of users. For the research, the researchers gathered 393 social media users between the ages of 17 and 29 who had experienced some form of negative effects of social media usage. These participants were then divided into three groups: a tutorial group which learned healthy social media habits, an abstinence group which stopped using social media completely, and a control group which continued their usual social media usage. The researchers tracked the social media activities of these groups for six weeks and assessed their mental health aspects.

The results of the study showed that the abstinence group saw a huge decline in their social media usage, while the tutorial group also became a little mindful and started using social media selectively. The results also showed that the abstinence and tutorial group spent less time scrolling on social media passively and stopped comparing themselves to others. The tutorial group saw significant changes while the abstinence group saw the biggest changes.

The researchers found that the results from each group proved to be effective in different mental health aspects. The tutorial group method is effective in reducing loneliness and FOMO, while the abstinence group method can help reduce anxiety and depression. On the other hand, the abstinence method wasn't effective for reducing loneliness, probably because it cut off social connection. This shows that completely cutting off social media may seem highly effective for various mental health issues, it can also lead to social isolation. The tutorial method helped participants use social media when it truly mattered, and they learned to notice when social media makes them feel good or bad, they started unfollowing or muting accounts that triggered negative feelings in them and they started engaging with their friends and family through comments and messages instead of mindless scrolling.

Image: DIW-Aigen

Read next: Keto and Low Carb Diet Proven to Reverse Several Chronic Diseases, Yet Big-Pharma Influence Suppresses Mainstream Awareness
by Arooj Ahmed via Digital Information World

Did the Biden Administration Order Big Tech to Censor AI Platforms? Republican Congressman Issues New Letter to Find Out

16 of America’s leading tech companies were issued letters, including the likes of Google and OpenAI, to determine whether or not the Biden administration had ordered censorship on AI tools.

The letter was issued by Republican Congressman Jim Jordan who demanded reports of past communications that might give rise to this alarming finding of the former president regarding censorship of lawful speech inside AI platforms.

We’ve seen Trump’s top tech advisors hit on the subject in the past, including how they would be getting to the bottom of this. This matter is the next phase inside the culture war between Silicon Valley and conservatives. Many held the opinion that silencing voices on various social media apps was unlawful to begin with. Now, the Trump administration wants to know if AI giants and other intermediary firms were involved in it or not.

Letters were released to the heads of OpenAI, Apple, and Google, where a report was highlighted that spoke about the Biden administration’s efforts to suppress speech inside AI.

The latest case also saw other tech companies like Amazon, IBM, Meta, Scale AI, Microsoft, Cohere, and more get asked for more details on this matter. The deadline is March 2,7, where all companies are asked to provide replies on this front.

What everyone is curious about is how one leading tech giant is missing from the list, and it’s Musk’s xAI. This obviously has a lot to do with the fact that Musk’s entry into politics and closeness with the president makes him immune to such investigations.

Many companies were anticipating such an investigation to come sometime soon and that’s why many leading AI firms have altered the ways AI chatbot assistants handle all sorts of politically sensitive questions.

During the year’s start, OpenAI shared how it was altering how it trains AI models to represent different perspectives and make sure ChatGPT was not censoring certain people’s views.

OpenAI denied that this was an attempt to appease the Trump admin but instead a means to double down on the firm’s core values. Meanwhile, Anthropic shared how its latest AI model won’t reply to fewer queries any longer. No matter how controversial the question might be, it would try its best to provide a nuanced reply.

Other firms were slower in the race to make changes to their AI systems and how they tackle political prompts. Remember. Google shared how its Gemini won’t be giving out replies to any political themed queries. Interestingly, even after the elections were over, the assistant does not consistently reply to political questions such as who is the head of state of America.

Some tech giant executives like Meta’s CEO appear to have added more fuel to the fire of conservative accusations regarding censorship at Silicon Valley. Zuckerberg mentioned that his firm was pressurized to silence certain kinds of content related to COVID-19 misinformation.

Image: DIW-Aigen

Read next: Apple’s Legal Battle Hearing with UK Government Against Access to User Data Heard Behind Closed Doors
by Dr. Hura Anwar via Digital Information World

Apple’s Legal Battle Hearing with UK Government Against Access to User Data Heard Behind Closed Doors

A hearing in the Apple legal battle with the British government regarding access to user data was recently held behind closed doors.

The news comes just one day after we saw privacy groups in the UK demand the proceedings be done more transparently as the public’s rights were at stake. Despite that, the press failed to get entry into the courtroom and was denied access.

The American tech giant rolled out an appeal against the decision made by the tribunal regarding the Home Office’s decision to demand encrypted data that’s found on Apple’s iCloud servers. Many media outlets, including the BBC, Computer Weekly, and the Guardian, made submissions for entry, but that did not go as planned and they were denied entry.

It was interesting to see Sir James Eadie, who usually represents the government in top cases, make an entry to the courtroom for this hearing.

The iPhone maker is fighting a tech capability notice rolled out under the nation’s Investigatory Powers Act. This forces organizations to help different law enforcement agencies in getting evidence.

This notice demands access to the company’s ADP service that focuses mostly on highly sensitive encrypted data belonging to users that is stored remotely across servers.

Apple shared how it was not in favor of the decision. Therefore, the order was challenged at the tribunal, which is designed to determine if the domestic challenge service act as per law or not. The news comes after the Cupertino firm chose to withdraw from the ADP in February this year.

Apple time and time again reiterates how it never designed a backdoor or any kind of master key for products and will never be following this kind of practice in the future as well.

The ADP makes use of E2E, which means only account holders get the chance for file decryption. Meanwhile, messaging services like iMessage and FaceTime continue to be E2E through default means.

The government’s hidden legal demands are commonly known as a tech capability notice or TCN. Therefore, recipients of this TCN can’t share the existence of this order until they’re provided with permission from the home secretary. The website also spoke about how hearings should be dealt with in private only when it’s deemed absolutely necessary to do so.

It also mentioned how there needs to be zero information disclosure that might be a threat to the country’s national security. Two days back, American lawmakers were seen calling on the tribunal to get rid of this cloak of secrecy related to the British government’s order and make Friday’s hearing public.

As per a recent report from Bloomberg, officials in the UK rolled out talks with American counterparts regarding this order. Therefore, the UK continues to assure America that it’s not getting any kind of blanket access. It only wishes for data related to serious criminal offenses or terrorist attacks, or matters linked to abuse.

Image: DIW-Aigen

Read next: Google All Set to Replace Google Assistant with Gemini This Year
by Dr. Hura Anwar via Digital Information World

Google All Set to Replace Google Assistant with Gemini This Year

Tech giant Google has confirmed that it is closing its doors on Google Assistant as that will be replaced with Gemini later on in 2025.

Google shared how the change would apply to all smartphones, tablets, vehicles, and devices that link to your phone like smartwatches and headphones. The same would be the case for television-linked devices.

The news is not surprising at all for most of us because when the tech giant rolled out Bard, it shared how it would bring forward Bard features to its Assistant. Then last year, we saw the company get rid of a host of features found on Google Assistant.

The company shared the news through a blog post where it mentioned how it was busy upgrading more users on cellphones to make way for this change. Later that same year, it mentioned how Assistant won’t be accessible on most devices for new installs through the mobile app store.

The company shared how it’s busy upgrading tablets, vehicles, and devices that link to your smartphone. It’s also keen on introducing the latest Google-based experience to all home devices such as televisions and displays.

The company posted how the change wouldn’t be for older devices like Android 9 and earlier phones that don’t have the right RAM specifics. The requirements for this device on Android entail the Gemini app being up for grabs on nearly all Android devices having 2 GB RAM and running the Android 10 system and beyond. Other than that, it includes Android tablets like Pixel Tablet featuring 2GB RAM or beyond and again having Android 10 and above.

The news is great as Gemini is really equipped to do a lot of tasks on Google’s behalf. So we don’t think anyone would mind the change. This change also signals how Gemini might be the future for Google in not only Android but also for Search.

Image: DIW-Aigen

Read next:

• What the Latest SEO Study Means for Your Business’s Online Visibility

• Global Smartphone Sales Flat in Q4 2024 as Android Faces Record Lows in India, US
by Dr. Hura Anwar via Digital Information World

Global Smartphone Sales Flat in Q4 2024 as Android Faces Record Lows in India, US

According to a new report by Counterpoint Research, there were little to no changes in global smartphone sales in Q4 2024, which led to only small changes in global OS shares. Android had 74% market share, the same as last year, but still faced some challenges. Even though Android was stable, it hit its lowest quarterly share in India and the US. Google and Motorola sales growth helped in the double-digit YoY declines among smaller brands in the US. The iPhone also has strong market stability with its sales only declining by 1%.

There was a 4% market share reported of Harmony OS, with unchanged YoY. In China, Harmony’s OS share increased by 19% mostly because of sales of new launches. For the fourth consecutive quarter, Harmony OS stayed ahead of iOS. On the other hand, the smartphone subsidies in China boosted local OEMs, which will probably result in further growth.

Read next: TikTok Leads Revenue and Downloads Despite US Ban Concerns, ChatGPT Surges, DeepSeek Joins Top Ten
by Arooj Ahmed via Digital Information World