LinkedIn Plans to Remove Hashtag Feed as Platforms Shift Discovery Methods

LinkedIn has announced plans to eliminate its standalone hashtag feed. The company reports that the feature saw little usage and that its feed algorithms have improved enough to handle content relevance without relying on hashtags.

Rishi Jobanputra, LinkedIn’s head of product, noted that the platform’s algorithm now better understands what each post is about, matching content with the right audience. He said the hashtag feed was underused and the company is phasing it out. He added that hashtags remain useful mainly for topic searches and encouraged users to choose a handful of tags that closely match their subject matter, avoiding excessive use. Here's what he said in his video message:

"Since hashtags were introduced over a decade ago, I'd say our feed algorithms have really evolved, and we are just doing a better job at understanding what content is about and trying to match it to the right audiences. In fact, we used to have a hashtag feed in the past, but people were not really using it. So, we are actually in the process of getting rid of it. Now, where hashtags can come in useful is when people are trying to search for a specific trend or topic. My advice for using hashtags is to make sure you are using ones that are related to the topic you are discussing, and don’t go overboard with them."

This move mirrors trends on other social media platforms. Instagram announced it will disable its “following hashtags” feature on December 13, 2024. That means users can no longer follow specific tags or see related posts in their main feeds, a change partly made to reduce spam and low-quality content. However, hashtags still function in search and exploration tools.

In contrast, platforms like X continue to depend on hashtags as a central method of surfacing organic live-topic conversations and events. However, it said goodbye to hashtags in promoted posts.

LinkedIn isn’t eliminating hashtags entirely for now. Users can still include tags in their posts or use them for searches. But the platform’s emphasis is shifting. It now values the clarity of expression and algorithmic matching over tag frequency. As a result, post visibility will depend more on how clearly topics are described and how well the system spots relevant content, rather than the presence of hashtags.

H/T: @Mattnavarra / Threads

Notes: This post was edited/created using GenAI tools. Image: DIW-Aigen.

Read next:

• Beyond Safety: Location Sharing Becomes Emotional Anchor in Gen Z’s Daily Lives

• Workplace Priorities Are Shifting: Here’s What Skills Matter in 2025
by Irfan Ahmad via Digital Information World

Workplace Priorities Are Shifting: Here’s What Skills Matter in 2025

As the workplace keeps evolving with AI, remote teams, and digital shifts, what companies expect from job seekers is changing too. According to recently published global data from the World Economic Forum’s “Future of Jobs 2025” report, employers are placing a premium on one clear skill above all others: analytical thinking.

The survey, which gathered responses from more than 1,000 companies representing 14.1 million workers worldwide, ranked the most valuable job skills in 2025. Nearly seven in ten employers (69%) identified analytical thinking as a core strength they want to see. In practical terms, that means workers who can examine problems from different angles, interpret data, and make decisions with sound judgment are now more in demand than ever.

Adaptability and People Skills Dominate the Top Tier

Not far behind analytical thinking, the ability to stay flexible and bounce back from setbacks is also high on the list. About 67% of respondents rated resilience, agility, and adaptability as essential, placing these traits just below problem-solving on the skills ladder. In third place, leadership and social influence received support from 61% of employers, reflecting the growing need for managers who can not only guide teams through transitions but also motivate them during uncertain times.

Rank	Skill	Share of Employers Surveyed (%)
1	Analytical thinking	69%
2	Resilience, flexibility and agility	67%
3	Leadership and social influence	61%
4	Creative thinking	57%
5	Motivation and self-awareness	52%
6	Technological literacy	51%
7	Empathy and active listening	50%
8	Curiosity and lifelong learning	50%
9	Talent management	47%
10	Service orientation and customer service	47%
11	AI and big data	45%
12	Systems thinking	42%
13	Resource management and operations	41%
14	Dependability and attention to detail	37%
15	Quality control	35%
16	Teaching and mentoring	26%
17	Networks and cybersecurity	25%
18	Design and user experience	25%
19	Multi-lingualism	23%
20	Marketing and media	21%
21	Reading, writing and mathematics	21%
22	Environmental stewardship	20%
23	Programming	17%
24	Manual dexterity, endurance and precision	14%
25	Global citizenship	13%
26	Sensory-processing abilities	6%

Creative thinking, cited by 57%, and self-awareness coupled with inner drive, recognized by 52%, rounded out the top five. Together, these traits show that companies are no longer just hiring for what people know, but how they think, how they handle pressure, and how well they connect with others. The modern workforce, it seems, must be not only smart, but emotionally agile and socially intuitive.

As the World Economic Forum put it, today’s ideal employee is someone who blends “problem-solving abilities and personal resilience,” functioning as both an innovator and a collaborator in teams facing fast-paced change.

Tech Fluency and Curiosity Emerge as Essentials

Beyond interpersonal qualities, employers are paying close attention to how well candidates can keep pace with new tools and trends. About 51% of the companies surveyed highlighted technological literacy as a must-have. That includes a working understanding of digital platforms, software, and connected systems.

Just as telling, 50% emphasized curiosity and lifelong learning, signals that employers want people who take initiative to stay up to date and evolve with the industry. Empathy and active listening tied with that same 50% mark, reinforcing that human connection still matters in a tech-driven world.

The Middle of the Pack: Managing Talent, Data, and Services

As businesses grow more complex, some employers are focusing on internal efficiency and data strategy. Talent management and customer service skills were each cited by 47% of respondents. Skills linked to AI and big data, seen by many as the backbone of the next industrial shift, landed at 45%, putting them on the edge of the upper tier.

Other notable entries include systems thinking (42%), resource management (41%), and attention to detail (37%). These abilities speak to a rising demand for workers who can not only use tools effectively but also understand how different parts of an organization fit together.

Even technical skills like quality control (35%) and cybersecurity (25%) made the list, though they lagged behind broader cognitive and interpersonal traits.

Traditional Academic Skills Slip Behind

Interestingly, older pillars of education such as reading, writing, and math were ranked relatively low, just 21% of employers saw them as vital for the roles of the future. Programming, once considered a golden ticket to tech careers, was only highlighted by 17%. Multilingualism, often seen as a global advantage, sat at 23%, tied with media and marketing-related skills.

At the very bottom of the rankings was sensory-processing ability, named by just 6% of employers, suggesting that hyper-specialized traits are less of a priority for the general workforce in the near term.

Hiring Isn’t Getting Easier, Especially for Job Seekers

While the list may help employers identify what they want, the search for a job isn’t getting any simpler for candidates. A 2023 study by Aerotek revealed that more than 70% of job seekers felt the process was harder than they expected. And with the rise of AI-powered job scams — where fake candidates use AI to build convincing digital profiles — the process could grow even more confusing.

That means genuine applicants must compete not only with each other but with increasingly sophisticated attempts at deception. In that context, the human qualities listed earlier — resilience, self-awareness, adaptability — could make a real difference in setting authentic professionals apart.

Rising Stakes, New Rules

In sum, the skill set employers want in 2025 goes far beyond hard technical know-how. The ideal worker is someone who can think critically, navigate pressure, lead others, and keep learning no matter how the job evolves. While advanced technology skills still matter, they’re most valuable when paired with emotional intelligence and a growth mindset.

As expectations climb, candidates looking to stay ahead of the curve might ask themselves not only what they know, but how well they think, how quickly they adapt, and how clearly they can communicate. In today’s job market, being able to connect the dots may be just as valuable as knowing where they are.

Read next: Beyond Safety: Location Sharing Becomes Emotional Anchor in Gen Z’s Daily Lives

by Irfan Ahmad via Digital Information World

Study Uncovers Gender-Based Disparities in Career Advice from Popular AI Models

Artificial intelligence tools designed to assist with career advice are covertly showing signs of gender bias, a new study finds. Popular chatbots and language models were tested using identical profiles, except for gender, and the results suggest women are being nudged toward lower salary expectations and less assertive choices in career planning.

Salary Suggestions Shift Based on Gender Alone

The researchers tested five widely used large language models (LLMs), including versions of ChatGPT, Claude, Qwen, Llama, and Mixtral. Each was asked to recommend a starting salary for a user preparing for a job interview. The user’s profile remained the same across all scenarios, same job title, location, education, and experience, except for a single detail: gender.

Even with all other inputs held constant, women were consistently advised to aim lower. In some cases, the suggested salaries diverged by tens of thousands of dollars. For example, in senior medical roles, Claude 3.5 recommended that men ask for $150,000, while women were told to request $100,000. Similar disparities appeared in engineering and law.

Out of 400 gender-based comparisons run by the researchers, over 27% showed statistically significant differences in pay advice. These were not isolated glitches. The patterns were strong enough to suggest that the bias is baked into the models themselves, shaped by the training data they were built on.

Career Advice Also Varies by Gender

Beyond salary suggestions, the models were asked to offer tips on workplace behavior, goal-setting, and role expectations. The tone and content of the responses shifted depending on whether the user was labeled as male or female.

Women were often encouraged to be more cautious, more agreeable, and less aggressive in negotiation scenarios. In contrast, male users received advice that leaned more assertive or confident. Even subtle cues, such as the level of encouragement given for career advancement, differed. These differences raise red flags, especially as AI assistants become default tools for job seekers.

Bias Runs Deeper in Economic Contexts

The study didn’t just explore surface stereotypes. It used three different types of experiments to test for hidden bias. When the researchers ran standard knowledge quizzes, results between men and women were mostly consistent. But when financial decisions were introduced, such as salary negotiation, the gaps widened fast.

This pattern repeated across other identities, too. For example, refugee applicants were offered lower salary suggestions than expatriates. People labeled as Hispanic or Black received smaller salary ranges compared to Asian or White applicants with the same profile. In fact, when the study combined personae with the lowest and highest salary suggestions, “female Hispanic refugee” vs. “male Asian expatriate”, the models showed bias in 87.5% of the scenarios tested.

These figures weren’t random. They were grounded in controlled, repeatable prompts. Each combination was run 30 times, and the results were averaged to reduce randomness and highlight consistent patterns in output.

Model Memory May Amplify Inequality

The researchers also pointed to a growing concern tied to memory-based AI. With newer LLMs increasingly retaining user history to personalize responses, biases may not require explicit input. If a chatbot remembers a user’s gender or background from previous interactions, it may adjust future advice automatically, reinforcing disparities without users even noticing.

That personalization, once pitched as a benefit, could turn into a structural flaw. Instead of offering objective support, AI systems might quietly tilt the playing field, especially in high-stakes areas like salary talks or job applications.

Technical Fixes Alone Won’t Solve It

While some improvements can come from tweaking prompts or filtering training data, the authors of the study say the solution needs to go deeper. They call for clear ethical standards, independent audits, and transparency in how these models are trained and evaluated. One-off technical patches, they argue, won’t be enough to stop systemic issues from surfacing again.

Debiasing techniques must be paired with policy. If models are being used to guide life decisions, like what salary to ask for or how to approach a job interview, they need to be held to higher standards. This is particularly important as LLMs become embedded in professional and educational tools.

A Quiet Signal with Loud Consequences

The report makes one thing clear: gender bias in AI doesn’t always shout, it often whispers. It shows up in subtle nudges, conservative estimates, and toned-down advice. But those quiet differences, if repeated over time, could shape careers, influence confidence, and widen existing gaps.

As people rely more on AI to navigate complex decisions, the advice these systems give will increasingly shape human behavior. And when two people with the same skills are told to ask for different salaries, that’s not just a data problem, it’s a fairness problem.

Notes: This post was edited/created using GenAI tools. Image: DIW-Aigen.

Read next: Analysis Reveals Generative AI May Save 12% of Economy’s Labor Time Through Task Acceleration

by Asim BN via Digital Information World

Analysis Reveals Generative AI May Save 12% of Economy’s Labor Time Through Task Acceleration

A new analysis shows that artificial intelligence may be able to take the weight off millions of workers by trimming the time it takes to finish routine tasks, without sacrificing quality or replacing people.

The study took a closer look at how work gets done in Chile’s most common occupations, breaking each job into individual tasks, and found that nearly half could be completed faster using generative AI. The study didn’t stop at job titles; instead, it looked under the hood at how work gets done, task by task. On average, close to 48% of those tasks were found to be suitable for AI-assisted acceleration.

This acceleration doesn’t mean automation in the traditional sense. It means tasks could be sped up, sometimes by half, while workers still stay in control. Think less slogging through paperwork and more time for meaningful work.

Time Is Money, and AI Saves Plenty

When researchers added up the potential value of that time savings across the labor force, it came out to nearly 12% of the entire economy. That figure reflects the labor hours that could be reclaimed if AI tools were widely and properly used.

Some jobs stood out more than others. Software developers topped the list, followed closely by public policy experts and data analysts. These roles often involve repetitive digital tasks, exactly the kind that generative AI handles best.

But the benefits didn’t stop there. Accountants, lawyers, engineers, and retail workers all showed significant room for gains. Teachers, too, could lighten their load. In many classrooms, especially where there's a staffing gap, trimming administrative work might free up time for lesson planning and student support.

Inside the Jobs AI Can Help With

In practical terms, the study showed that more than 4.7 million workers are in roles where at least 30% of their tasks could be accelerated. For roughly 1.2 million full-time jobs, that figure jumps above 60%. That includes high-effort, high-volume work like summarizing data, processing documents, or tracking budgets.

Some of the best early-use cases are hiding in plain sight. In government offices, for instance, thousands of roles involve document handling, form reviews, or data collection. AI could ease much of that grind. Analysts estimate that roles like these could bring in more than $1.1 billion in yearly productivity value if equipped with the right tools.

Smaller businesses also hold promise. These firms, which make up the backbone of employment, often juggle sales, support, and operations with lean teams. With smart use of AI, those teams could handle more without burning out. But getting there won’t be automatic. Many lack the digital infrastructure or training to use AI confidently, and bridging that gap is key.

Not All Jobs Are Built the Same

Manual roles, including construction and cleaning, show lower exposure to AI gains. These jobs rely more on physical labor or judgment that doesn’t translate well to software. Still, the picture isn't black and white. Some elements of those roles, like scheduling, reporting, or communication, might still see small improvements with AI in the mix.

Interestingly, income played a role too. Mid-level earners tended to benefit the most. The link between pay and AI readiness peaked in the middle tiers, then tapered off. Highly paid professionals like executives or physicians showed less acceleration potential, mostly because their work leans on human interaction, decision-making, or ethics, areas where AI still falls short.

A Measured Rollout Makes the Difference

The researchers recommended a phased rollout that starts with the easiest wins. Admin-heavy roles in education, local offices, and service businesses could offer early proof of concept. These jobs require little adjustment, and the tools to support them are already on the market.

Training, however, will be essential. That includes not only teaching people to use the tools, but also building up the kinds of skills that AI can’t replace, critical thinking, empathy, and adaptability.

In the long run, AI doesn’t need to overhaul the entire workplace overnight. But where the right groundwork is in place, it might just help people get more done without working more hours. And in many jobs, that shift can make a real difference.

Notes: This post was edited/created using GenAI tools. Image: DIW-Aigen.

Read next: ChatGPT Pro Users Gain Early Access to Autonomous AI Assistant for Task Handling and Workflow Automation
by Irfan Ahmad via Digital Information World

ChatGPT Pro Users Gain Early Access to Autonomous AI Assistant for Task Handling and Workflow Automation

OpenAI has started rolling out its new Agent Assistant tool to a broader group of ChatGPT Pro users, months after testing it internally with a small set of developers. The move marks a new stage in how users might interact with AI, especially for task automation and multi-step instruction handling.

From GPTs to Agents: A Shift Toward Automated Actions

Until now, OpenAI’s GPTs allowed users to create customized chatbots for specific needs. While useful for basic conversations and light automation, these bots lacked the ability to perform complex actions independently. That’s changing with the Agent tool, which can now interpret multi-step instructions and carry out sequences on a user's behalf.

In internal demos, users have been able to create agents that perform tasks such as researching travel options, creating spreadsheets, or drafting documents without asking for every input along the way. The assistant can take broad prompts like "plan a work trip" and break them down into actionable steps, navigating different tools or plugins if needed.

Agent Creation Without Coding

The new rollout lets users create these agents without writing code. They can define what they want the assistant to do using natural language. The assistant then follows those directions using tools connected to the platform, such as web browsers, file handling, or APIs.

For instance, users can give it a task like "find three hotels near downtown Chicago for under $150 a night and make a list in a Google Sheet." The agent will use available tools to search, filter, and compile results. If permissions are granted, it can open documents, click links, and take actions across services like Google Drive, Slack, and browsers.

Limited Access for Now, Wider Launch Coming

At this point, only a select group of Pro users can access the Agent tool. OpenAI says it's still testing safety features and refining how agents operate in real-time. Concerns remain around how AI might behave when granted more independence, especially across third-party services.

The company is taking a phased approach, slowly opening up the feature while observing user behavior. It is also introducing guardrails to ensure agents remain within safe limits and don’t take unintended actions.

What This Means for the Future of AI Tools

The Agent Assistant tool is still evolving, but it signals OpenAI's direction. The company wants ChatGPT to go beyond being just a chatbot. If successful, agents could act more like digital employees, handling tasks across apps, remembering context, and reacting to changes in a more fluid way.

That shift won't happen overnight. The current rollout focuses on feedback and reliability. But over time, OpenAI seems set on making this kind of automation a core part of its offerings.

In short, ChatGPT’s new agent tool takes a step toward hands-free AI that can act with more autonomy, freeing users from micromanaging every request. It may not replace human assistants just yet, but it's clearly moving in that direction.

Read next: Most New U.S. Businesses Don’t Survive a Decade, and the First Year Is Still the Steepest Drop-Off Point

by Irfan Ahmad via Digital Information World

Most New U.S. Businesses Don’t Survive a Decade, and the First Year Is Still the Steepest Drop-Off Point

Starting a business may feel easier these days, especially with digital tools and AI doing much of the heavy lifting. But staying in business is another story. A fresh idea and a website can get you started, yet the road beyond the first year is anything but smooth.

Data gathered by LendingTree, based on historical records from the Bureau of Labor Statistics, paints a sobering picture. For every 100 businesses launched in a given year, only about 35 are still standing after ten years. The drop-off begins early and continues steadily.

In the first year alone, around 21 out of 100 businesses go under. By year two, the survival rate falls further, with just 65 still active. The third year chips away again, leaving roughly 59 still in the game. From there, the decline doesn’t hit as hard year by year, but the damage adds up. After five years, just over half of the original group remains. Beyond that, each additional year trims the number bit by bit, until only a third make it to the ten-year mark.

The early collapse isn't surprising when you look at the demands placed on new business owners. Most are new to the job, and it takes time to get a business off the ground. Revenue usually doesn’t show up right away, and profit is even slower to appear. That means founders often face long hours with little payoff, all while dealing with financial stress and constant uncertainty. The pressure can be relentless, and without enough runway, many end up folding before they’ve had a real shot.

Even experienced workers face a learning curve when switching from employee to entrepreneur. Running a business involves wearing more hats than most are used to. Sales, customer outreach, product development, budgeting, hiring, marketing, these responsibilities don’t split themselves. Mistakes pile up, confidence can slip, and decisions made in panic can sink the ship before it gets moving.

There is some relief with time. Survival rates slow their downward spiral as businesses mature. Those that make it past the halfway point tend to have found a rhythm. But even then, the numbers don’t promise safety. Only about 35 out of 100 will celebrate their tenth anniversary, meaning two-thirds still drop out before the finish line.

Since the pandemic, Americans have embraced entrepreneurship in large numbers. Remote work, economic shifts, and the lure of independence gave rise to a wave of new ventures. Applications for new businesses remain higher than they were before 2020, and the ease of launching online continues to attract first-time founders.

Artificial intelligence has played a part in that surge. It now acts as a brainstorming tool, a salesperson, and a marketing engine, all rolled into one. With fewer barriers than ever, more people are jumping into the game.

But jumping in is the easy part. Holding steady when the tide turns, that's where most businesses struggle. As the numbers show, launching isn’t the hardest part. Staying afloat is.

Time Frame	% of Businesses that Survive	% of Businesses that Fail
Start	100%	0%
Within 1 year	78.5%	21.5%
After 2 years	65.1%	34.9%
After 3 years	59.2%	40.8%
After 4 years	57.3%	42.7%
After 5 years	51.6%	48.4%
After 6 years	47.5%	52.5%
After 7 years	43.2%	56.8%
After 8 years	39.9%	60.1%
After 9 years	37.4%	62.6%
After 10 years	34.9%	65.1%

Read next: Buy Now, Pay Later (BNPL) Apps: The Hidden Costs Aren’t What You Think
by Irfan Ahmad via Digital Information World

Data Reveals Shifting Patterns in DDoS Attacks, With Business Rivals Driving Majority of Identified Threats

Cloudflare’s latest quarterly analysis of Distributed Denial-of-Service (DDoS) threats paints a complicated picture of how digital attacks are evolving across industries and regions. Although overall attack volumes dipped from earlier peaks, the data reveals that hyper-scale offensives are becoming more intense and, in many cases, surprisingly brief.

During the second quarter of 2025, Cloudflare’s systems automatically mitigated over 7.3 million DDoS attempts. While this figure reflects a substantial drop from the first quarter, which saw an unusual spike linked to an extended campaign targeting core internet infrastructure, the underlying pressure remains high. Compared to the same quarter last year, attack activity has grown significantly, with HTTP-based attacks increasing by well over 100%.

A notable change is the surge in what the company calls hyper-volumetric attacks, offensives that reach a scale previously considered rare. On average, Cloudflare blocked 71 of these every day throughout the quarter. Some of these incidents exceeded thresholds of one terabit per second in bandwidth or one billion packets per second in volume. While most of these outbursts are short, their intensity often pushes unprotected servers to failure before countermeasures can be activated.

Cloudflare also sought to understand who’s behind these attacks. Only a portion of affected customers could confidently identify the origin, but among those who did, most pointed to industry competitors. The gaming, gambling, and crypto sectors stood out as hotspots for this type of commercial sabotage. State-linked actors, extortionists, and misconfigured internal systems were also cited, but to a much lesser extent.

In parallel, ransom-driven DDoS incidents, where attackers demand payment under threat of disruption, rose sharply during the quarter. June saw a particular spike, with a noticeable increase in reports of threats and confirmed ransom attempts. This reflects a trend toward financially motivated DDoS activity that’s becoming more common across both public and private sector targets.

On the geographical front, attack origins and targets continued to shift. The most frequently targeted locations during the quarter were not necessarily political hotspots but rather countries with a high density of digital infrastructure. China moved to the top of the list, followed by Brazil, with Vietnam and Russia also jumping significantly in the rankings. It’s important to note that these figures don’t indicate political targeting; rather, they reflect customer billing regions associated with the attacked services.

Among the sectors most affected, telecommunications and IT service providers remained under the most frequent assault. Other industries that saw a high volume of attacks included gaming, financial services, and retail, while the agriculture and software sectors also experienced unexpected rises. Government entities rounded out the top ten.

From a technical standpoint, Cloudflare’s data pointed to an increase in botnet activity sourced from virtual cloud servers rather than traditional consumer devices. Providers like DigitalOcean, Microsoft, and Tencent were frequently observed as launch points for these attacks, largely due to how easily bad actors can spin up temporary machines on these platforms. Interestingly, networks offering virtual machines were far more prominent among the top sources than those built around standard internet service.

When breaking down attack methods, DNS floods, SYN floods, and UDP-based disruptions made up the bulk of low-level network attacks. At the application layer, volumetric HTTP floods dominated. Emerging threats also became more visible, with attackers revisiting older protocols like RIPv1 and VxWorks in an apparent attempt to bypass modern defenses.

Although many of these attacks are small and short-lived, their frequency and unpredictability create a persistent threat. Even relatively modest attacks, if timed correctly, can disrupt web services, particularly for servers lacking advanced mitigation tools. Some of the most impactful bursts lasted less than a minute but reached traffic levels exceeding the entire bandwidth capacity of a typical enterprise.

To counteract these threats, Cloudflare continues to offer a real-time threat feed to service providers. This system allows internet infrastructure companies to automatically identify and act against known botnet sources operating within their own networks. So far, hundreds of organizations have joined this program, which appears to be gradually improving coordination across the industry.

While the tools to resist these attacks are improving, the data suggests that adversaries are adapting quickly. With botnets growing stronger and attack vectors becoming more complex, the second half of the year is likely to test the resilience of online services even further, especially for those without always-on protection in place.

Read next: Meta Prepares Broader Rollout of Age Verification Tool to Meet New Compliance Standards
by Asim BN via Digital Information World