AI has resulted in a massive traffic surge on the internet. Big Box retailers are backing away from self check-out. And did Sam Altman get fired because AI could do simple math?
 |
 |
 |
These and more top tech stories on Hashtag Trending
I’m your host Jim Love, CIO of IT World Canada and Tech News Day in the US.
Dozens of news outlets have reported that OpenAI was reportedly working on an advanced artificial intelligence model, called Q* or “Q-Star in OpenAI documents,” which raised significant safety concerns among its researchers. This model, capable of solving relatively simple math problems, represents a notable advancement in AI development.
Why? It sounds ridiculous. After all, computers can do math. Yes they can, but we forget that every instruction that makes that work is programmed in by a human.
We’re talking about a computer intelligence that can reason through mathematical problems where the methods are not pre-programmed. This is, for many, early evidence of a machine intelligence that can learn by itself without the need for human intervention.
Andrew Rogoyski of the University of Surrey’s Institute for People-Centred AI commented on the significance of a large language model (LLM) capable of solving mathematical problems, calling it a major step forward in AI’s analytical capabilities.
Altman himself has hinted at this in a Wall Street Journal interview where he said that it was possible that in the future, AI wouldn’t require massive amounts of data from the internet and other sources. Some have taken this to mean that OpenAI had developed a way for AI to not only learn, but to actually teach other models by creating virtual data.
This would be a massive breakthrough.
Which is why there are stories about researchers at OpenAI being alarmed by Q*’s capabilities that they wrote to the board of directors, warning that it could pose a threat to humanity.
Some are speculating that it was these developments that led to the general panic and the dismissal of CEO Sam Altman.
All this has reignited discussions on the pace at which companies like OpenAI are advancing toward Artificial General Intelligence (AGI), a level of AI that can perform tasks at or above human intelligence levels and potentially operate beyond human control.
These developments raise critical questions about the ethical and safety aspects of rapidly advancing AI technologies, particularly in the realm of AGI, and OpenAI’s commitment to developing “safe and beneficial artificial general intelligence for the benefit of humanity.”
And am I the only one who is thinking – Q – isn’t that the name of the all-knowing sometimes evil super intelligence on Star Trek?
Sources include: The Guardian
Shout out to Sujan Sarkar at writerbuddy.ai who sent me a study that looked at the AI industry’s growth from September 2022 to August 2023 and looking at over 3,000 AI tools.
50 AI tools got 24 billion visits in that 12 month period. Of the 24 billion total visits during this period, ChatGPT was the clear leader. It attracted 14 billion visits, which is 60 per cent of the top 50 AI tool traffic.
ChatGPT, Character AI and Google Bard showed big growth. But not everyone was a winner – Craiyon, MidJourney, and Quillbot had significant traffic declines during the same period.
A couple of other interesting stats. There was an almost 70 – 30 split of males to females in terms of visits. And this might be a great reason for developing AI apps for mobile – 63 per cent accessed AI tools on mobile devices.
We’ll post a link to the full study with the show notes at itworldcanada.com/podcasts.
A recent trend has emerged among some big-box retailers, including Walmart and a huge Canadian retailer Canadian Tire. Some of their stores are backing away from self-checkout machines in favour of traditional cashier-based checkouts. This comes from a CBC News report, and marks a significant reversal from the initial embrace of self-checkout technology that promised to cut labor costs and speed up transactions.
Dwayne Ouelette, the owner of a Canadian Tire store in North Bay, Ontario, removed self-checkout machines to enhance customer service. He stated, “I’m not comfortable using them and I don’t think some of my customers are comfortable [either].” This sentiment reflects a broader dissatisfaction among customers and store owners alike with self-checkout systems.
Retail adviser David Ian Gray noted that while self-checkout was initially seen as a technological advancement to improve customer experience, it has led to various issues, including technical glitches and increased theft. The lack of supervision at these self-service checkouts has contributed to a rise in theft, a concern echoed across the retail industry.
A study funded by the industry revealed that 23 per cent of store losses could be attributed to theft and customer error at self-checkouts. Additionally, a survey by LendingTree found that 15 per cent of Americans admitted to stealing at self-checkout machines.
Customers have expressed their preference for this new format, valuing the personal interaction with cashiers. The report indicates a growing trend where shoppers and retailers are increasingly favouring traditional checkout experiences over automated ones.
Sources include: CBC News
A recent investigation by Blackwing Intelligence, conducted in collaboration with Microsoft’s Offensive Research and Security Engineering group, has uncovered significant vulnerabilities in Windows Hello’s fingerprint authentication system. Presented at Microsoft’s BlueHat conference, the research demonstrated how the fingerprint authentication on certain laptops could be bypassed, potentially allowing unauthorized access.
The vulnerabilities were found in three specific laptop models: a Dell Inspiron 15, a Lenovo ThinkPad T14, and a Microsoft Surface Pro 8/X, each using fingerprint sensors from different manufacturers (Goodix, Synaptics, and ELAN, respectively).
The researchers pointed out that these security flaws stem from issues in the communication between the software and hardware, rather than Windows Hello or fingerprint technology itself.
The bypass techniques involved manipulating the fingerprint sensor’s data. In one method, an attacker could use a Linux boot to rewrite the sensor’s data, allowing unauthorized access. Notably, this attack could be executed while the computer is on, using a man-in-the-middle (MITM) device, making it more challenging to prevent. Other methods would allow some machines to be bypassed with simple devices even when booting up.
Despite these vulnerabilities, Blackwing Intelligence clarified that full-disk encryption and BIOS passwords could still provide effective security measures, as exploitation is less feasible if the machine cannot boot to the point where fingerprint authentication is available.
The researchers urged manufacturers to use Secure Device Connection Protocol (SDCP) and properly integrate sensor chips with Windows to address these security gaps. They plan to provide more detailed information about the vulnerabilities they discovered in the future, underscoring the ongoing importance of cybersecurity vigilance in the face of evolving threats.
Sources include: The Register
Security researchers have discovered two new zero-day vulnerabilities being actively exploited to incorporate routers and video recorders into a hostile botnet. These vulnerabilities, previously unknown to both manufacturers and the security community, allow remote execution of malicious code on devices using default administrative credentials. The attackers have been leveraging these vulnerabilities to infect devices with Mirai, a powerful botnet software, to conduct distributed denial-of-service (DDoS) attacks.
The vulnerabilities are present in specific models of network video recorders and a wireless LAN router intended for hotels and residential applications, produced by a Japan-based manufacturer. The affected devices were found to have security flaws in the communication between their software and hardware. Akamai has reported these vulnerabilities to the manufacturers, with one confirming that security patches will be released next month.
The exploitation of these vulnerabilities involves command injection, requiring the attacker to authenticate themselves using the credentials configured in the vulnerable device. But these devices are notorious for having weak or easily guessed passwords.
A quick Internet scan by Akamai revealed at least 7,000 vulnerable devices, but the actual number could be higher.
Mirai, the botnet software used in these attacks, gained notoriety in 2016 for its massive DDoS attack capabilities. The current Mirai strain, primarily an older version known as JenX, has been modified and shows similarities to other Mirai variants.
This discovery underscores the ongoing threat posed by IoT botnets and the critical importance of cybersecurity vigilance in protecting against such sophisticated attacks on internet connected devices.
Sources include: Ars Technica
And that’s the top tech news for today.
Hashtag Trending goes to air 5 days a week with a special weekend interview show we call “the Weekend Edition.”
You can get us anywhere you get audio podcasts and there is a copy of the show notes at itworldcanada.com/podcasts
And I’ll cover cybersecurity stories that I think are of general interest, but you can keep up todate on cybersecurity with our podcast featuring security journalist Howard Solomon and called CybersecurityToday. It’s often rated as one of North America’s top 10 tech podcasts.
I’m your host Jim Love. Have a Fabulous Friday and I’m off to AWS’s re:Invent for the week and the dulcet tones of James Roy will take you through the week. Talk to you when I’m back and watch for my stories posted through the week on IT World Canada and TechNewsDay in the US.