Fact-checking AI content... Is it possible?

Fact-checking AI content... Is it possible?

Everywhere I turn these days, there is another comment or opinion on artificial intelligence or AI. Digging a little deeper, it would appear the concerns are for what comes after GPT-4.

I listened to part of a Joe Rogan episode that was completely generated by AI, including the voices of Joe and his guest. When it comes to this kind of thing, if it is flagged as such, then I don't see the problem. If however someone is using it to cause damage of any kind then this needs to be checked. Some countries are already starting to block even the latest version of ChatGPT, which makes little sense to me. Like any information you get online, it should be checked or verified as correct. I don't mean the fake fact-checkers you see around the place. I mean actual checking of source material where possible. The vast majority of the so-called fact-checkers are biased or bought and paid for by special interest groups.

- In related news, OpenAI, the creators of ChatGPT, are not currently training GPT-5, rather they are looking at ways to optimise what they already have and increase the capabilities of the current model. What has happened and was predicted by some, is the increase in ChatGPT-driven spam bots giving group admins at sites like Reddit increasing problems. Based on the latest Elon Musk interview with Tucker Carlson, OpenAI should change its name to ClosedAI as they now charge for services and its code and product are behind a private wall. Elon was one of the founders of OpenAI but he left when the direction of the organisation started to change in areas he was not happy with.

- In the same interview, Elon suggested he was looking at creating a competitor to ChatGPT that would be truth-seeking and fact-based. This product would sit in the middle of the political spectrum and ideally not be influenced in any way other than by the facts and data. Good luck Mr Musk.

- It's been a while, but you may remember the little helicopter called Ingenuity that was flying around Mars. It was designed to fly five times but it recently clocked up its 50th mission. To date, the longest flight was 169.5sec and the longest distance 704m. It has hit speeds of 6.5m per second and so far has flown about 89 minutes in total covering 11.5km. Nasa has been underestimating the capability of its spacecraft for decades now with examples like Voyager lasting far longer than anyone thought. This is what real engineering means.

- A story from theregister.com by Simon Sharwood caught my eye this week. It was about a student working on a thesis and continuing an earlier work. The previous student's work was required for the thesis. The previous student agreed. But after waiting and waiting, the next student decided to take things into his own hands. Writing a script to gain root access to the information, he soon had what he needed to proceed with the task. But he ended up being vilified for this. I'm not advocating hacking but I applaud the student's drive to get the work done. It also highlights just how porous systems were and are. Remember, if it's online then eventually someone will get hold of it.

- Microsoft has been trialling ads since November 2022 in the Windows 11 Start Menu, at this stage for its own products and services. Initially, it was a few users but it has steadily grown. According to Reddit users, they are not impressed. Signing in with your Microsoft local user account triggers this "benefit" but you do also get free access to Microsoft 365 and the cloud. This is all part of the move to the operating system as a service rather than you owning it outright.

- Today's AI is typically a product story that comes from China. China's State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing (LIESMARS) apparently allowed artificial intelligence to control the Qimingxing 1 satellite. Qimingxing 1 is described as an experimental craft for use by Wuhan University students. The AI "decided" to focus on an Indian army unit involved in actions with the Chinese, and on a Japanese port that sometimes has American naval ships visit. There is no mystery here, the AI was directed to look at locations of strategic interest to China. The AI did not unilaterally decide to do these things and pretending otherwise is disingenuous.

- Google is now facing a combined lawsuit from 17 US States on the grounds of unfair trade practices through its advertising. The suit claims a monopoly on the buying and selling of digital ads through its auction process. The claim is also against the forced use of Google's tools and the manipulation of auction outcomes. None of this is a surprise to anyone in the industry, the surprise is that some States are finally taking Google to task over it. The aim may be to break up Google into smaller pieces but that remains to be seen.

James Hein is an IT professional with over 30 years' standing. You can contact him at jclhein@gmail.com.

Do you like the content of this article?

TPIPP to phase out coal-fired power generation by 2026

TPI Polene Power will continue to allocate its 5-year capital spending of B14bn to replace its coal usage with renewable energy and boost its power generation capacity.


6 more officers embroiled in lorry 'sticker' scandal

The investigation into "sticker bribes" for illegally overloaded lorries has incriminated another six officers from the Highway Police Division (HPD), according to Pol Maj Gen Jaroonkiat Pankaew, the Counter Corruption Division (CCD) commander.


Ratchanok, Supanida, mixed pair prevail

Ratchanok Intanon, Supanida Katethong and mixed doubles pair Supak Jomkoh and Supissara Paewsampran moved into the last 16 round of the US$850,000 Singapore Open on Tuesday.