Tech News: Concerns about AI on the Rise

how the AI cookie crumbles. Before we get into all that, one story that I missed on Tuesday was an update to Microsoft's quest to acquire video game giant Activision Blizzard. So quick recap. Microsoft announced its plan to acquire Activision Blizzard, which was going through a really rough PR patch to put it lightly way back in early twenty twenty two. Now, since then, these companies have encountered resistance among various regulators around the world, primarily in the United States and the

United Kingdom. While some of those knots have been loosened, at least one that was really looming was the US Federal Trade Commission, or FTC and its request for an injunction to block the acquisition. Now, to be clear, this move would have been like a temporary hold on the deal, while the FTC mounts a more substantial legal challenge, But on Tuesday, a federal judge decided against that request for injunction and the deal would be allowed to move forward.

Microsoft wants to close this deal by July eighteenth, so it's coming up right soon. The FDC is now appealing to the Ninth Circuit Court of Appeals. Unsurprisingly, reps for Microsoft and Activision Blizzards say the deal will actually be really good for consumers as well as for competition. I

would humbly suggest that consolidation is is rarely good for competition. Still, the federal judge says that the FTC failed to show it has adequate justification to object to the deal in the first place, that the FTC's arguments aren't really legally convincing. Now I've not read all these arguments, but I can imagine that the FTC is having a tough time arguing their point, because it's pretty difficult to show how a deal is going to have negative impact on competition before

the deal happens. We usually only recognize these things in hindsight. Either way, the announcement means we are a little bit closer to Activision Blizzard joining the Microsoft family. Now, we've got a lot to talk about with AI today, and first up, I figure I should mention the ongoing labor

strikes in Hollywood. For a few months now, the Writer's Guild of America has been on strike, and today sag AFTRA, which is an actors' union, is poised to join them after the negotiating committee failed to reach an agreement with movie studio producers and the committee came to a unanimous vote to support a strike. Now, there are a lot of different issues at play here, but one of those

is the role of AI in Hollywood. So understandably, there are concerns that studios could choose to lean on AI rather than say, a human writing staff to create entertainment. So writers are worried that they're going to be called in to punch up an AI generated script and then receive a lower payout because they didn't actually write the first draft. They were just hired to punch up an already existing script. This could end up being a lot of work because AI is not necessarily great at writing

such scripts. Meanwhile, actors are worried about their likeness and their voices as AI gets increasingly better at creating digital duplicates and how does that impact their ability to make a living. Keep in mind that most of the people in Hollywood aren't your mega celebrities who have multi million dollar houses in Beverly Hills. Most of them are working professionals. You know, they're going gig to gig and making what they can and you know, saving and all that kind

of stuff. They're not all millionaires. Based off an article and Deadline that seems to confirm every stereotype of producers being greedy and manipulative jerkfaces, the studios are eager for

any option that maximizes profit and shareholder value. That article in Deadline essentially said that the producer's strategy is to wait until October before coming back to the negotiating table with the WGA the writers because by that point, the writers will be desperate because they'll be, you know, having trouble paying rent and mortgages. So it's it's pretty much pure evil. When you read that article, it's kind of hard to come to any other conclusion. It has really

galvanized the labor movement in Hollywood right now. Now, it's quite possible that this means we're going to be in a real dry spell as far as movies, TV and streaming content goes at least any content that would be coming out of Hollywood and that that could last for a little while. And I think this battle is one that's needed. Of course, I am biased because I'm still fuming that my old place of employment decided to eschew human writers in favor of AI. But y'all don't need

to hear me go off on that again. If you did miss it, just listen to a couple of recent episodes where I talk about what happened at houstuffworks dot com. Now it turns out that sacking employees in favor of AI doesn't play well to the general public. CEO Summit Shaw found that out on Twitter. He posted on Twitter that his company quote had to lay off ninety percent of our support team because of this AI chatbot. Tough, yes, necessary,

absolutely end quote. He claimed that this AI chatbot outperformed his customer support staff, so he had no choice but to replace human beings with artificial intelligence. He said that this chatbot could resolve problems in a fraction of the time that it took human workers, saying that it was a difference of minutes with AI versus hours with humans, and that this saved on customer support costs by up to eighty five percent. Now Folks on Twitter descended like

piranha on Shaw's message. Some argued that the layoffs really had more to do with a failing business model and not an indication that AI is actually superior to human staff, essentially saying Shaw was using AI as a shield to

guard against perceptions that his company was doing poorly. Shaw has not backed down from his claims and says it might be that he's on the more aggressive side and that you know, talking about and taking action as something that perhaps other business leaders are reluctant to do, but that this is where we're all headed fun times, because I don't see how that ends up being a win in the long run. The Clerkson Law Firm has proposed a class action loss suit against Alphabet, Google, and DeepMind.

Alphabet is the parent company to those other two companies. The law firm accuses the companies of rifling through well pretty much everything that's ever been online and all for

the purposes of training AI language models. The lawsuit argues that Google, without permission, used personal, professional, copyrighted, and private works, including stuff like private email, all to train AI and that this represents a huge breach of trust and privacy, and that those afflicted, which is pretty much anyone who has ever had any of their information posted online, ever,

it gives them the right to deserve justice. Google's own privacy policy, which updated on July first, says it may collect data that is quote publicly available online end quote. But the law firm argues this is drastic overreach, that just because something is publicly available, that doesn't mean it's free for use for whatever purpose, and that seems reasonable to me. Now. According to one of the plaintiffs in the lawsuit, there is evidence that google bard was trained

on copyrighted works, including this person's works. And if you use the correct prompts with google Bard, you could essentially get the works presented back to you. So like you could ask what's the beginning of chapter three of such and such and it would recite it to you. And that means that there could be a workaround for people who want to access a writer's work, but you know, they don't want to have to pay for it. And I get it. The Internet has trained us to expect

stuff for free. We've been kind of conditioned to expect that. However, like I said at the beginning of this episode, if writers can't make a living writing, then writing kind of stops. So the approach ultimately becomes self destructive, and we're not done with AI yet. Business Insider has an article about how folks who created applications and processes that are built on top of GPT four, which is the most recent build of open AI's large language model, have started to

run into some issues recently. The article references users on Twitter who have said that the model has become less useful and described it as being dumber or lazier than it had been previously. Problems include making careless mistakes, which makes me think back to the days when I was

taking math courses. Sometimes I would forget something simple like a negative sign, and so I would get an entire problem wrong, even though you know, step by step I was doing things correctly, but because I was careless or lacked attention to detail, I got the wrong answer, and thus you know that was a valuable lesson in paying

more attention. Well, apparently GPT four has some of those same issues, where it will make some of these careless mistakes like drop a bracket, which in code can really just lead to a massive a bug or failure in the program. Now, some have said the tool is also less likely to quote unquote remember between prompts. That means it gets to be harder to use GPT four to build stuff, because at some point or another in the

building process it forgets what it was doing. The trade off on this is that GPT four is now faster than it had been at launch. So the developer community is growing to suspect that open ai made some changes that speeds things up, but at the cost of accuracy and performance. Open Ai so far has declined to speak about the perceived problems. All right, we've got a lot more stories to cover before we get to that. Let's

take a quick break. We're back and I think it can wrap this up in one more segment of store. So starting off, yesterday, US Congress representatives released a report accusing three major tax preparation companies, that being H and R, Block, Tax Act, and Tax Slayer of sharing sensitive private customer data with Meta, you know, Facebook's owner, and that Meta was using this information to what else, improve its targeted advertising strategy, and that these actions weren't taken with the

knowledge or consent of customers. And here's where I remind folks outside the United States that over here in the US, tax preparation can get really complicated, and for a lot of people it falls way outside their comfort zone. So tons of folks depend upon third parties to prepare their taxes. I understand this concept is strange in some other countries where a lot of that work is already done by

the government on behalf of its citizens. Pretty sure that wouldn't fly here in the States because there are too many people here have a deep distrust in the government, particularly when it comes to handling something like your personal finances. I can't entirely blame them for that, By the way, our government has done a lot of boneheaded stuff anyway.

According to the report, these companies shared information like income amounts, how much people were expecting in refunds, how much tax they owed the government, the names of their dependence and more. In addition, one of those companies, Taxpayer, was also accused of having shared similar information with Google. The congressional representatives

are calling for a full investigation into the matter. There's been some talk that this information was designed in such a way as to not make it quite as personally identifiable. But I think that's all rubbish because we have seen time and again it takes very few points of data for you to narrow in and figure out who that

data belongs to. We've seen it time and again. You only need a few points of data to be able to weed out the vast majority of folks and zoom in on the people it could possibly be, which means you could quote unquote anonymize data all you like and you could still make it possible to figure out who

that data belongs to. Anyway, Once again, I feel like I need to go on the broken record route and call for the need for comprehensive data security and privacy laws here in the US, because of course companies are going to engage in this type of behavior if there's even just a little ambiguity over whether or not it's legal. So it needs to be spelled out in black and white what is and is not legal, and what the

consequences are for breaking the law Before that happens. We're going to see stuff like this over and over again. Microsoft revealed that Chinese hackers were able to access Microsoft email accounts belonging to two dozen government agencies and a massive data breach. The agencies include both US and European departments. The attackers managed to gain access earlier this year May at the latest, but maybe even earlier than that. Microsoft

was unaware of the intrusion until mid June. According to security experts, the attacks were extremely sophisticated and obviously precisely targeted. The vulnerability appears to have been in Microsoft's cloud services, which highlights one of the big challenges with cloud computing.

Organizations end up having to hand over at least partial control to a third party and in trust that the third party is going to protect the systems and information that are nested in the cloud on behalf of the customer. The alternative is to keep all your systems on premises, but as systems grow more sophisticated and complex, and as organizations become more distributed, the physical requirements of sticking strictly with on prem becomes untenable. So there is a real

need for things like cloud computing. But it also shows that cloud computing requires massive investments in security, which Microsoft has done. It's just this was a case where bad actors were able to exploit a vulnerability and it went unnoticed for at least a few months or at least two months. Amazon has filed a challenge to the EU's Digital Services Act or DSA that's scheduled to become a

fully operational act death Star style by August twenty fifth. Now, this Act is part of the EU's efforts to create rules and restrictions and regulations that apply to big tech companies. If the EU determines the tech company counts as a quote unquote very large online platform or VLOP, then that company has to meet certain obligations in order to comply with these new regulations. So among those criteria is the requirement to police disinformation and hate speech on the platform.

Amazon's not super keen on being held accountable for doing that and wants to dispute its classification as a VLOP. Now, considering Amazon is a leading cloud computing company, it seems like it's a pretty steep hill to climb to say that we're not a VLOP. But then Amazon reps have a few salient points to make. Amazon is primarily known as an online retail company. They say, we're not a

search engine, at least not in the traditional sense. We're not a social network, even though there are some social network functionalities built into Amazon, such as user reviews. And further, Amazon argues that large retailers in the EU are not being classified as vlops, So why is Amazon being held to a standard when other large retailers in the EU

are not. The reps say that while Amazon agrees that the company bears responsibility to make sure that the products it offers on its marketplace are legal, it should not fall into the classification of a very large online platform because it is fundament different from platforms like Facebook or Google. EU reps haven't directly addressed Amazon's statements, but have indicated that all vlops will need to take steps to comply with the law. However, the process for doing that will

be different depending upon the nature of the platform. So a marketplace will do this in a very different way than a social network, for example. And finally, I have a recommended article for y'all today. It's from Lucas Ropek of Gizmodo. The article is titled a proposed law would force Internet companies to spy on their users for the DEA. So this really applies to people living in the United States.

It all stems from a very real problem how do you crack down on illegal and deadly drug trade online without also creating a surveillance state. So I recommend you check out that article. It's really well written, it's extensive. It goes into detail about the actual nature of the problem, why this is something that we do need to address, and why the proposed solution could potentially be a really

bad one. Again, the title of that article is a proposed law would force Internet companies to spy on their users for the dea over at gizmoto, And just full disclosure, I have no connection to Gizmoto. I do not know Lucas Robek. I just thought this was a very good article that folks should read. And that is it for the news for Thursday, July thirteenth, twenty twenty three. I hope you are all well, stay safe out there, and I'll talk to you again really soon. Tech Stuff is

an iHeartRadio production. For more podcasts from iHeartRadio, visit the iHeartRadio app, Apple Podcasts, or wherever you listen to your favorite shows.