HomeLatest ThreadsGreatest ThreadsForums & GroupsMy SubscriptionsMy Posts
DU Home » Latest Threads » Emrys » Journal
Page: 1 2 3 4 5 6 ... 66 Next »


Profile Information

Gender: Do not display
Current location: Scotland
Member since: Sun Sep 6, 2009, 11:57 PM
Number of posts: 6,546

Journal Archives

OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic

ChatGPT was hailed as one of 2022’s most impressive technological innovations upon its release last November. The powerful artificial intelligence (AI) chatbot can generate text on almost any topic or theme, from a Shakespearean sonnet reimagined in the style of Megan Thee Stallion, to complex mathematical theorems described in language a 5 year old can understand. Within a week, it had more than a million users.

ChatGPT’s creator, OpenAI, is now reportedly in talks with investors to raise funds at a $29 billion valuation, including a potential $10 billion investment by Microsoft. That would make OpenAI, which was founded in San Francisco in 2015 with the aim of building superintelligent machines, one of the world’s most valuable AI companies.

But the success story is not one of Silicon Valley genius alone. In its quest to make ChatGPT less toxic, OpenAI used outsourced Kenyan laborers earning less than $2 per hour, a TIME investigation has found.

The work was vital for OpenAI. ChatGPT’s predecessor, GPT-3, had already shown an impressive ability to string sentences together. But it was a difficult sell, as the app was also prone to blurting out violent, sexist and racist remarks. This is because the AI had been trained on hundreds of billions of words scraped from the internet—a vast repository of human language. That huge training dataset was the reason for GPT-3’s impressive linguistic capabilities, but was also perhaps its biggest curse. Since parts of the internet are replete with toxicity and bias, there was no easy way of purging those sections of the training data. Even a team of hundreds of humans would have taken decades to trawl through the enormous dataset manually. It was only by building an additional AI-powered safety mechanism that OpenAI would be able to rein in that harm, producing a chatbot suitable for everyday use.


OK. This is one reason why Twitter needs an edit function.

I initially thought it was the nephew who cried.

I'm glad to find out I was wrong.

As well as the methods explained in the first two replies,

you can also hotlink some images hosted on other sites. I say "some" as different sites have different attitudes to hotlinking, as they may be sensitive to people effectively stealing some of their bandwidth.

As long as the image ends in the suffix .jpg, .jpeg or .png it should display as long as the host site doesn't have hotlinking blocked. Some media sites carry images which can't be hotlinked in this way as their URLs don't end in any of those suffixes. As mentioned elsewhere in the replies, you need to use the post preview function to check whether the image will display on DU.

Twitter images can also be hotlinked, but it's a little more involved. If you see a Twitter image you want to post, say you're using Firefox, you need to left-click on the image so it displays in isolation, then right-click on it and a variety of choices will appear. Select Copy image link then you can paste the URL into the DU posting or reply box. Choosing a random image I found, you'll end up with a URL something like this:


You need to delete everything from the ? onwards (including the ? ) and replace it with .jpg. It should then display like so:

Be aware that some on DU are sensitive to Twitter content, especially since Musk took over. They're unlikely to object to a simple image as nobody will necessarily be aware of where it's from.

Some GIFs will display on DU, but it's hit or miss whether you can find one that ends with the suffix .gif - some GIF hosting sites have sharing tools which will help.

You can also post YouTube content. You just need to copy the URL from your browser's navigation box and paste it, and it should embed and play when clicked on.

The fact Musk has developed this vision of Twitter as having "sides",

and that he identifies with one "side" over another and sees the other as some sort of an enemy says it all.

Nobody can serve as a successful moderator of a community with that dynamic going on, and he'll have an enormously tough time ahead trying to forge anything like a functional Twitter if he carries on like that (which he will, because I don't think he's capable of changing). Any successor will also face a huge task because Musk's poisoned the well.

With the majority of Twitter's active users seemingly utterly fed up (or worse) with Musk and paid blue-tickers rapidly gaining pariah status and being blocked by some on sight, it's not exactly a well thought out recruitment campaign for the paying customers he's fixated on getting.

Divisions obviously exist on Twitter, always have. But they're not as simple as that bipolar framing.

Musk pissed off a number of users he'd probably assume to be on his "side" with that stupid and shortlived scheme to ban outlinks to rival platforms. Opposition came from all quarters - and that's what you'd expect given the libertarian populist rhetoric he's been spouting.

It's like his commitment to free speech - the more you pay, the freer it is, unless he takes something you post personally, in which case you'll be forced to go take your speech elsewhere and be free with it.

It's been fun watching him trying to rationalize his decisive defeat in the poll about whether he should quit as CEO or not. Kim Dotcom (yes, he's still around) insisted it was down to liberal bot activity, and pointed out that Musk now had data from the poll to identify those bots and eliminate them. Musk's response was the reply "Interesting."

This on a platform where he was gloating a few weeks ago at having conquered the bot problem, posting a GIF of him gurning and gesticulating over a bot grave!

What he did when he posted that poll was fire the starting pistol for a competition between those who wanted him to remain as CEO and those who didn't. Both sides of opinion had access to similar tools to spread the word and mobilize votes. It just so happens that one side is more numerous on Twitter and has better networks (including DU). If the result came as a surprise to him, then he should take it as a wake up call. His reaction since doesn't convince me that the poll was a clever ruse to let him back away from Twitter gracefully. He's moaning that nobody who could fulfil the role wants to take it on. He's trapped himself, with nobody involved in any of his three main businesses happy with what he's doing, and like quicksand, the more he struggles at the moment, the more trapped he becomes.

So now you have his acolytes wandering around Twitter moaning after the event that they didn't know about the poll, that 12 hours was too short for it to run, that it was held at the wrong time of day on the wrong day of the week, etc. etc. Some have even been driven to frame the poll as a brilliant trap set by Musk to weed out users who're disloyal to him! Whether that means they think he's going to kick off those millions who voted to fire him, I've no idea, but some of them are crazy enough.

Most of Musk's angst at the moment seems to be to do with money (despite the fact he's a nominal multibillionaire), and more notably, Twitter's looming bankruptcy, which he's discussed openly with some on Twitter itself. It's almost like he has no pals or trusted peers to turn to in his everyday life.

It's incredible, but less and less surprising as we watch him flailing in action, that he didn't realize from the start that Twitter's been very good for many things over the years. It's just that making money hasn't been one of them.

I'm not clear what you mean by Twitter "subscribers".

If you mean active users, that 2 million figure is way wrong. These were the figures last summer:

The figures below are based on Twitter’s advertising audience data for July 2022.

Note: our regions follow the United Nations Geoscheme. Click here to learn which countries belong to each region.

Number of active Twitter users in Northern America: 92.1 million

Number of active Twitter users in Central America: 19.1 million

Number of active Twitter users in the Caribbean: 3.3 million

Number of active Twitter users in Southern America: 43.1 million

Number of active Twitter users in Western Europe: 30.6 million

Number of active Twitter users in Northern Europe: 27.2 million

Number of active Twitter users in Southern Europe: 20.4 million

Number of active Twitter users in Eastern Europe: 8.2 million

Number of active Twitter users in Northern Africa: 8.0 million

Number of active Twitter users in Western Africa: 6.3 million

Number of active Twitter users in Middle Africa: 551 thousand

Number of active Twitter users in Eastern Africa: 4.0 million

Number of active Twitter users in Southern Africa: 3.4 million

Number of active Twitter users in Western Asia: 44.7 million

Number of active Twitter users in Central Asia: 377 thousand

Number of active Twitter users in Southern Asia: 30.9 million

Number of active Twitter users in Southeastern Asia: 57.6 million

Number of active Twitter users in Eastern Asia: 81.6 million

Number of active Twitter users in Oceania: 4.9 million


Despite the exodus after Musk took over, the overall figures won't have changed much.

Legal action launched against Michelle Mone-linked PPE Medpro

THE UK Government has initiated legal action against PPE Medpro, the firm linked to Tory peer Michelle Mone at the centre of a row over Covid contracts issued through the Conservatives’ illegal “VIP lane”.

The Department of Health and Social Care said it had issued breach of contract proceedings over the 2020 deal on the supply of sterile gowns, with the BBC reporting that it is seeking to recoup £122 million, plus costs.

PPE Medpro said the civil case would be “rigorously defended” and accused the DHSC of a “cynical attempt to recover money from suppliers” who acted in good faith.

A statement issued by the firm said: “PPE Medpro will demonstrate to the courts that we supplied our gowns to the correct specification, on time and at a highly competitive price.

“The case will also show the utter incompetence of DHSC to correctly procure and specify PPE during the emergency procurement period. This will be the real legacy of the court case and it will be played out in the public arena for all to see.”


At last a prosecution, though not the one many were hoping to see involving Mone herself.

Ballsy stuff from Medpro's spokeperson, which neatly sidesteps the issue of how the PPE contracts were awarded to government cronies in the "VIP lane" in the first place and came to be so vastly overpriced, though that may end up being the subject of further legal action.

Earlier OPs about the Mone affair:

Revealed: Tory peer Michelle Mone secretly received 29m from 'VIP lane' PPE firm

Revealed: Second firm pushed by Michelle Mone was secret entity of husband's office

Michelle Mone's businessman husband faces jail if found guilty of Spanish tax charge

I'm willing to lay a modest wager that at least one of the replies to Musk

will include this 2014 pic of him and convicted child trafficker Ghislaine Maxwell:

If he sees it, Musk may deflect and come out with his usual excuse that Maxwell "photobombed" him.

That would be all very well, but ...

Ghislaine Maxwell reportedly asked Elon Musk to destroy the internet in the famous photo of the two that the billionaire claims was a photobomb
The photo was taken in 2014 at the Vanity Fair Oscar Party in California, but it didn't make a splash on social media until 2020, when Musk took to Twitter to say he didn't know Maxwell "at all."

"She photo-bombed me once at a Vanity Fair party several years ago," he tweeted in July 2020. "Real question is why VF invited her in the first place."

But, the story seems slightly more complicated, according to a recent report from The New York Times. When the photograph was taken, Maxwell asked Musk if there was a way to delete information about oneself off the internet, the Times reported on Tuesday. Maxwell also told the Tesla CEO she wanted him to destroy the internet, a Vanity Fair worker who was standing next to the pair told The Times.
In 2020, Insider reported that Epstein introduced Musk's younger brother, Kimbal Musk, to a woman in his entourage in order to get close to the Tesla CEO. Kimbal and the woman dated from 2011 to 2012, and Epstein had received a tour of Elon's company SpaceX in 2012, sources told Insider. At the time, SpaceX denied Epstein received a tour and Musk also said on Twitter that Epstein never toured SpaceX "to the best of our knowledge." 


Something something about glass houses and stones

This is yet another thing Musk hasn't thought through, and the countermeasure is laughably simple:


James Felton
Sorry nobody saw your post, it didn’t appeal to bootlickers

Corvid Stan Account

I like how this just incentivises us to block all paid checkmarks.

The reply is followed by ones from a number of people who already have a policy of blocking accounts with paid blue checkmarks on sight.

Oh, and what is the point of all these posts telling those of us who use Twitter what we MUST do?

I don't see it persuading anyone, and they're just a magnet for the already persuaded to chat among themselves and sometimes argue with the occasional interloper who doesn't agree.

Michelle Mone's businessman husband faces jail if found guilty of Spanish tax charge

Prosecutors in Spain confirmed they were seeking custodial sentences for seven British businessmen, including Doug Barrowman, ahead of their trial. All seven deny the charges
Businessman Doug Barrowman, 57, has been charged with crimes that include corporate tax evasion and could be jailed for more than five-and-a-half years if convicted.

Prosecutors in Spain confirmed they were seeking custodial sentences for seven British businessmen, including Mr Barrowman, ahead of their trial. All seven deny the charges.
The indictment relates to claims Barrowman and his co-accused benefitted from a “fraudulent” invoice.

Prosecutors allege it was created to evade tax and take millions of pounds out of a Spanish company to put in their pockets. The company went bankrupt four years later, costing 200 workers at a cable factory near Santander their jobs.


The case is now due to be heard next May. It's separate to the legal issues discussed in the earlier OPs below, but who knows what dots may be joined and whether Barrowman will be at liberty to attend the Spanish case in person if legal action is pursued against him in the UK?

Revealed: Tory peer Michelle Mone secretly received 29m from 'VIP lane' PPE firm

Revealed: Second firm pushed by Michelle Mone was secret entity of husband's office

I could handle reading less *about* Twitter and more based on the *information* it carries.

That includes people moaning about Twitter and those of us who continue to use it.

I understand that some think it important to keep up with Musk's latest meltdowns. I see far more about Musk here than I do on Twitter itself, but nobody forces me to read more than the post titles.

Twitter's at least two-edged, like most things in life (even DU). It's an unbelievable international resource, and it's long been deeply flawed, like anything that involves a large number of people.

Some people seem to think its problems began when Musk took over. They must forget how long it took for Jack Dorsey's team to take action on hate speech like antisemitism and ban Trump, for instance, and revolting content has been a constant irritant which you can largely avoid if you use the filtering tools provided and don't go looking for the worst of it.

Musk has deepened many of those problems, but they've always been something you have to find ways to step around if you want to make Twitter useful. At the moment, you can still do that.
Go to Page: 1 2 3 4 5 6 ... 66 Next »