Facebook’s parent Meta enters the AI fight with a surprising one-two punch, companies are already replacing workers with ChatGPT and remote work paranoia.


It’s Hashtag Trending for Monday, February 27th and I’m your host Jim Love, bringing you the top tech news stories for today.

On Friday, Meta, the parent of social media giant Facebook announced that it was entering the AI competition with a large language model that it claims will outperform OpenAI’s GPT-3.

If Meta’s announcement is to be believed, their new AI, called LLaMA, has two huge differences from OpenAI’s ChatGPT.  

First, Meta’s offering is ten times smaller.  

Second, Meta trained its AI using publicly available datasets, such as Common Crawl, Wikipedia, and C4, which makes the sourcing of the training much more transparent and allows the possibility that Meta’s model could be offered as open source.  And in a tweet issued by a project member Guillame Lample, it appears that making the model open source is a real possibility.

Lample states: Unlike Chinchilla, PaLM or GPT-3, we only use datasets publicly available, making our work compatible with open-sourcing and reproducible, while most existing models relay on data which is either not publicly available on undocumented.” 

The key to this breakthrough appears to be in what is termed parameter size. Parameters are the variables that a machine-learning model uses to make its predictions based on input data. Larger models typically are more capable of delivering much more complex answers. But since more parameters take exponentially more computing resources, if an AI could generate the same results with fewer parameters, it is capable of running with far less resources.  

Anyone who has tried to investigate ChatGPT and found the “server busy” message, has had direct experience of why a more efficient model can lead to greater usage.

Plus, a model with a reputed 10X efficiency advantage opens the way for ChatGPT style AI on devices such as PCs and possibly even smartphones. 

A copy of the link to Meta’s paper announcing their new model can be found in the text version of this podcast at ITWorldCanada.com

Sources for this include an article in ArsTechnica

Employers are already replacing workers with ChatGPT

Resumebuilder.com, an online app for resume development, surveyed 1,000 business leaders who already use or are planning to use ChatGPT. Nearly half of the companies have already implemented it, and of that roughly ½ of the companies who have adopted it, say they were already using it to replace workers.  

Resumebuilder’s Chief Career Advisor Stacie Haller said, “Since this new technology is just ramping up in the workplace, workers need to surely be thinking about how it may affect the responsibilities of their current job. The results of this survey show that employers are looking to streamline some job responsibilities using ChatGPT.

According to the survey results, some of the ways that companies are using ChatGPT include 66 per cent for writing code, 57 per cent for customer support and 52 per cent for meeting summaries and other documents. 

77 per cent of companies say they are using ChatGPT to help write job descriptions, 66 per cent to draft interview question and 65 per cent to respond to applications.

And in a news release from ResumeBuilder.com, the company reported that companies were extremely impressed with the results to date.  “Fifty-five per cent say the quality of the work produced by ChatGPT is ‘excellent’ and 34 per cent say it is ‘very good.’”

Haller noted that the “economic model for using ChatGPT is evolving.” But of the companies surveyed who were using it, nearly all said they’ve saved money, with 48 per cent saying they’ve saved $50,000 and 11 per cent saying they’ve saved more than $100,000. 

No wonder that 93 per cent said they plan to expand their use of ChatGPT.  A further note in from the survey – 90 per cent of executives say ChatGPT experience is beneficial for job seekers – provided of course, that it hasn’t already been used replaced their job.

Productivity paranoia?

Another survey, this one from Envoy.com claims that although 94 per cent of employees believe their managers trust them to do work from anywhere, only 24 per cent of those same employees trust their co-workers who work from home.

Larry Gadea, CEO and founder of Envoy said,
“Less visibility in the office is skewing perceptions and seeding a distrust of everyone and everything, starting from leadership and trickling down,” “Now, managers realize that the relationships and trust employees develop in person are critical to getting work done right, especially in today’s environment,”

There’s a significant gap between the responses of Gen Z, Millenials and Baby Boomers as well. Only 57 per cent of Gen Z “feel strongly they have the manager’s trust.” That’s compared with 71 per cent of Millenials and 77 per cent of Boomers.

But despite these differences, they are universally suspicious of their co-workers. Only 31 per cent of Gen Z, 23 per cent of Millenials, and 17 per cent of Boomers trust their colleagues to do work when working outside the office.

That distrust percentage varies radically based on whether the employees themselves work remotely or full time in the office. While 34 per cent of hybrid employees trust their colleagues, and only 10 per cent of those who work full time in the office feel the same way.

Almost all (98 per cent) felt that certain work is better done in the office. Training and onboarding, problem solving and brainstorming were favoured as in office by one half or more of the sample. Curiously, conflict management was only viewed by 37 per cent as more effective in the office and only 35 per cent thought that the office was best for meeting people from outside the organization.

One of the most telling statistics was in terms of the recent round of layoffs is that 92 per cent said that being seen at the office improves their job security.

Sources include: TechRepublic and envoy.com

And in more related news from what is the fastest growing app of all time, hackers are exploiting the popularity of ChatGPT to swindle unsuspecting users. 

ITWorldCanada.com’s security reporter, Howard Solomon reported a number of weeks ago that there were a number of applications in the Google Play Store which purported to be ChatGPT additions, but which actually contained malware or could potentially steal data or load malware to a user’s phone or PC.  According to a tweet by Dominic Alvieri, the number of those potentially malicious apps keeps growing,  Alvieri, , so devoted to security that he has reportedy deleted his own Facebook account –  reported that there were 1,453 domains registered this week including the words ChatGPT

One website, according to Alvieri, called chat-gpt-pc.online presented itself as a way to download ChatGPT as a local application for Windows. Alvieri found that the download would inject users with a malware called RedLine which stole information from the user’s web browsers. If, for instance, a user has passwords or credit card in information stored in their browser, this application could pull that information and send it to the hackers.

A report from cybersecurity firm Cyble discovered more than 50 fake ChatGPT apps. In addition to stealing data, one of these, called ChatGPT1 provides no AI services, but does covertly subscribe its victims to numerous paid services in an SMS billing fraud scheme.

Until such time as reliable listings of AI programs become available, the only reliable source of ChatGPT is at the OpenAI website or through the facility in the Bing web brower .

Despite layoffs, the average salary for tech workers in the United States remains in excess of six figures according to a leading salary database provider, Dice.com 

Dice’s 2023 Tech Salary report reports that “the average tech salary increased 2.3 per cent year over year to $111,000 (compared to $108,000) in 2022. Following a year of strong salary growth for tech professionals, the continued upward trend in average tech salary shows that tech skills and the professionals that provide them can still command high compensation because of sustained demand.

The top cities in terms of salaries had significant gains including Tampa, with a year over year increase of 22 per cent and Portland at 16 per cent.

Also, according to Dice’s analysis, six figure plus salaries for high tech workers are no longer limited to the coasts. Non-traditional tech hubs including Columbus, saw average tech salaries increase of 16 per cent year over year, and Phoenix saw salaries grew 26 per cent.

All of the top 25 cities for tech compensation boast an average salary over $100,000, and these locations are spread throughout the country.  For more information you can go to Dice.com/Salary Trends.

Source: Business Insider

And that’s the top tech news stories for today.  Hashtag Trending goes to air five days a week with daily news and we have a special weekend edition, bringing you an interview with an expert on some area of technology that’s making the news.

Follow us on Apple, Google, Spotify and anywhere you get your podcasts. You can get us on your smart speaker and instructions for that can be found on our podcast page at www.itworldcanada.com/podcasts where you can also find links mentioned in this podcast. 

If you are interested in cybersecurity, why not check out our sister podcast Cybersecurity Today. 

As always, we are glad to hear from you – a shout out to a couple of people who wrote me over the weekend and thanks for your comments. As always, I’ll take the time to respond personally to each of these and we will be working to incorporate your great suggestions. Your ideas and your comments help us serve your information needs better. Keep ‘em coming.

Thanks. I’m your host Jim Love – Let’s make it a great Monday!


The post Hashtag Trending Feb. 27th- Meta enters the AI battle; companies replace workers with ChatGPT; study reveals remote work paranoia first appeared on IT World Canada.

Leave a Reply