Tech Brew Ride Home - Thu. 04/20 – Elon’s Having A Heck Of A 4/20

Episode Date: April 20, 2023

Are Elon and Microsoft fully at war at this point? Does Coinbase already have one foot out the country at this point? Is OpenAI careening toward a run-in with GDPR? And, Stability AI’s big play to compete with Chat GPT. Sponsors: Leadership.OregonState.edu/cic Grammarly.com/techmeme Links: Microsoft drops Twitter from its advertising platform (Mashable) Elon Musk Threatens to Sue Microsoft After it Drops Twitter From Ad Platform (PCMag) Coinbase gets Bermuda license, plans to launch offshore exchange in coming weeks (FortuneCrypto) First EU-Wide Crypto Regulations Clear Final Parliament Vote (Bloomberg) OpenAI’s hunger for data is coming back to bite it (MIT Technology Review) Google to deploy generative AI to create sophisticated ad campaigns (Financial Times) Stability AI releases ChatGPT-like language models (TechCrunch) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the Tech Meme right home for Thursday, April 20th. The day Elon Musk probably won't be able to resist making a million dumb dad jokes about. I'm Brian McCullough today.

Are Elon and Microsoft fully at war at this point? Does Coinbase already have one foot out the country at this point? Is Open AI careening toward a run-in with GDPR and Stability AIs big play to compete with ChatGPT? Here's what you miss today in the world of tech. Looks like Twitter's revenue is going to take another hit. Microsoft plans to drop Twitter support. port from its ad platform on April 25th. Users of that platform won't be able to access or manage Twitter accounts through that tool. And here's what that means. Quoting Mashable. The Microsoft

Advertising feature previously allowed advertisers to manage their social media accounts on various platforms in one place. Users could respond to tweets and DMs along with messages received on Facebook, Instagram, and LinkedIn. While Microsoft's social media service was provided for free to advertisers, it was prominently featured in Microsoft Advertising's Digital Marketing Center dashboard. It worked alongside the platform's social and search paid advertising tools, which helped businesses run and manage their paid ad campaigns on platforms like Google Ads, Facebook and Instagram, and Microsoft's Search Advertising. Companies that use Microsoft advertising will still be able to manage and create content

for Facebook, Instagram, and LinkedIn through the platform, just as they were able to before. Microsoft made more than $12 billion in digital advertising revenue last year from ads that would be created, managed, and run through this advertising platform. Quote, starting April 25th, 20203, Smart Campaigns with Multi-Platform will no longer support Twitter, Microsoft said. A similar email has begun to go out to Microsoft advertising users stating that Digital Marketing Center will no longer support Twitter starting on April 25th. From that date, users will no longer be able to access their Twitter account through its

Digital Marketing Center's social media management tool, according to Microsoft. Users will also no longer be able to schedule, create, or manage tweets or tweet drafts. In addition, users won't be able to view their past tweets and engagement on the Microsoft advertising platform, end quote. So what I've heard is that this is all about those Twitter API changes. Basically, it was already barely profitable for Microsoft to provide this third-party platform access with Twitter, and that was under the old pricing. But now it's basically impossible for Microsoft to plug Twitter into this dashboard because the API calls are so expensive. So if you're a big brand advertiser, managing your tweets and your presence across

various platforms and social, you can still manage your Instagram stuff here, your LinkedIn ads, etc. The one big platform you'll have to manage elsewhere will now be Twitter. Now soon after this news dropped, well, this happened. Responding to a tweet from Twitter Daily News discussing Microsoft refusing to pay the API fees, quoting PCMag, they trained illegally using Twitter data. Lawsuit time, Musk said in a tweet referring to Microsoft News, Musk didn't provide any evidence for the illegal training, although programs such as ChatGBTGBT were trained on public internet data such as Wikipedia articles. However, Musk made his comment right as rival social media platform Reddit announced plans to charge for API access, citing how Reddit user-generated data is being used to train AI models. Musk also has a beef with Microsoft, which has become the major partner for OpenAI, the company behind ChatGPT. Although Musk helped found OpenAI, he's since been blasting the San Francisco-based lab as a quote, closed source, maximum profit company, effectively controlled by Microsoft, end quote. Elon responded to another tweet, which said,

I'm curious if you, Elon, have a long-term plan here. I understand charging for the API, but I think one thing that had been great about Twitter in the past is its ability to function across the Internet. In many cases, this move is killing traffic to Twitter itself from outside sources. For instance, embedding tweets elsewhere, which normally would drive new users to Twitter, is cut off in some cases, to which Elon replied, quote, I'm open to ideas, but ripping off the Twitter database, demonetizing it, removing ads, and then selling our data to others isn't a winning solution, end quote. Remember that recent story we did about Coinbase, maybe considering all options, including leaving the country.

If it can't get regulatory clarity from the U.S., well, quoting the block, Coinbase could launch an offshore derivatives exchange as soon as next week after the crypto exchange received a license to operate in Bermuda. The company obtained a license from the Bermuda Monetary Authority, according to a company blog post published on Wednesday. Coinbase lauded Bermuda as, quote, a highly respected and experienced financial regulator in its announcement. Coinbase had been exploring, launching an offshore platform to trade perpetual swaps tied to cryptocurrencies, the block reported last month. That expansion could come as soon as next week, a source familiar told the block. Fortune first reported the news. Bermuda was one of the first financial centers to pass comprehensive digital assets,

regulation in 2018, and its regulatory environment is long known for a high level of rigor, transparency, compliance, and cooperation, Coinbase said in its blog post, the company did not immediately respond to a request for comment. Perpetual swaps, a type of future and popular product in the crypto space, are expected to be among the offerings. They provide a more capital-efficient way for traders to make bets on the underlying crypto market, end quote. Did someone say regulatory clarity? Parliament in the European Union approved markets in crypto assets, or Micah, regulation for the crypto industry, with rules for stable coins set to apply from July of next year, quoting Bloomberg. European Financial Services Commissioner Mariet McGuinness said Wednesday that she expects the

legislation to come into force in July after it's formally approved by the bloc's 27 member states. Specific requirements will take effect progressively with rules governing stablecoins, for example, set to apply from July 24. Micah, in development for three years, has been welcomed by crypto executives as an alternative to the U.S. approach for policing the sector through enforcement actions. Yet critics say the law is outdated even before taking effect, as it would do little to prevent several of crypto's recent high-profile blowups, and there are already calls for updates. Once implemented, Micah will require any company offering crypto-related services in the EU to gain registration in one of the bloc's member states, which then allows them to operate across

the entire block. The European Banking Authority and the European Securities and Markets Authority will be in charge of making sure crypto platforms comply with the rules, including having adequate risk management and governance processes to avoid another FTX-style collapse. A unified EU-wide regulatory framework will likely make the block more attractive to digital asset companies and put pressure on other jurisdictions to follow suit, said Alisa DiCaprio, chief economist at Enterprise blockchain firm R3. Europe has been taking market share in crypto venture, capital investment from the U.S. data from Pitchbook show. It would be a surprise of other jurisdictions like the U.S. aren't quick to follow suit and further accelerate their crypto regulatory efforts,

DeCaprio said, end quote. Did someone say European regulation? Open AI, which may have trained its models on people's data without consent, might be facing legal strikes in the EU, which has strict privacy laws and is conducting probes as we speak. Quoting MIT technology Review. Open AI has just over a week to comply with European data protection laws following a temporary ban in Italy and a slew of investigations in other EU countries. If it fails, it could face hefty fines, be forced to delete data, or even be banned. But experts have told MIT Technology Review that it will be next to impossible for Open AI to comply with the rules. That's because of the way that data used to train its AI models has been collected by hoovering up content off the internet. In AI development,

the dominant paradigm is that the more training data, the better. OpenAI's GBT2 model had a data set consisting of 40 gigabytes of text. GPT3, which chat GPT is based on, was trained on 570 gigabytes of data. OpenAI has not shared how big the dataset is for its latest model, GPT4. But that hunger for larger models is now coming back to bite the company. In the past few weeks, several Western data protection authorities have started investigations into how OpenAI collects and processes the data-powering chat GPT. They believe it has scraped people's personal data such as names or email addresses and used it without their consent. If OpenAI cannot convince the authorities its data use practices are legal, it could be banned in specific countries or even the entire European

Union. They could also face hefty fines and might even be forced to delete models and the data used to train them, says Alexis Lauter, an AI executive in the French Data Protection Agency CNIL. OpenAI's violations are so flagrant that it's likely that this case will end up in the Court of Justice of the European Union, the EU's highest court, says Lillian Edwards, an internet law professor at Newcastle University, it could take years before we see an answer to the question posed by the Italian data regulator. In addition to being more transparent about its data practices, open AI will have to show it is using one of two possible legal ways to collect training data for its algorithms, consent or what is known as legitimate interest. It seems unlikely that OpenAI will be able to

argue that it gained people's consent when it scraped their data, that leaves it with the argument that it had a legitimate interest in doing so. This will likely require the company to make a convincing case to regulators about how essential chat GPT really is to justify data collection without consent, says Edwards, end quote. One of the first use cases mentioned for this generative AI stuff was to create ads, like given things like stable diffusion's ability to gin up pictures and chat GPT's ability to spin up text. Imagine the AB testing revolution that would be possible for ads where you could constantly be iterating ad copy and content

to get the best results in real time. Well, the Financial Times has seen an internal presentation that suggests Google plans to do just that, to deploy generative AI in their ads soon, allowing advertisers to supply creative material, which would then be what they call remixed in order to create campaigns. quote, Google already uses AI in its advertising business to create simple prompts that encourage users to buy products. However, the integration of its latest generative AI, which also powers

its Bard Chatbot, means it will be able to produce far more sophisticated campaigns resembling those created by marketing agencies. According to the presentation, advertisers can supply creative contents such as imagery, video, and text relating to a particular campaign. The AI will then remix this material to generate ads based on the audience it aims to reach, as well as other goals such as sales targets. One person familiar with Google's presentation said they were worried the tool could spread misinformation because text produced by AI chatbots can confidently state falsehoods. It is optimized to convert new customers and has no idea what the truth is, the person said.

Google told the Financial Times it planned to put firm guardreels in place to prevent such errors known as hallucinations when it rolls out its new generative AI features in the coming months, end quote. And finally today, speaking of Stability AI, that startup behind Stable DeFest, Fusion, which is that tool that creates those amazing photos you see around the internet. Guess what? They want in on the chatbot stuff too. Say hello to StableLM, its first suite of instruction fine-tuned LLMs, starting with 3 billion and 7 billion parameter flavors, but with 15 billion and 65 billion versions soon to follow, according TechCrunch, called Stable

LM and Available in Alpha on GitHub and Hugging Face, a platform for hosting AI models and code. Stability AI says that the models can generate both code and text and, quote, demonstrate how small and efficient models can deliver high performance with appropriate training. Language models will form the backbone of our digital economy, and we want everyone to have a voice in their design. The Stability AI team wrote in a blog post on the company's site. The models were trained on a dataset called The Pile, a mix of internet scraped text samples from websites including PubMed, Stack Exchange, and Wikipedia. But Stability AI claimed it created a custom training set that expands the size of the standard pile by 3x.

Stability AI didn't say in the blog post whether the stable LM models suffer from the same limitations as others, namely a tendency to generate toxic responses to certain prompts and hallucinate, i.e. make-up facts. But given that the pile contains profane, lewd, and otherwise fairly abrasive language, it wouldn't be surprising if that were the case. Still, the stable LM models seem fairly capable in terms of what they can accomplish, particularly the fine-tuned versions included in the alpha release. Tuned using a Stanford developed technique called Alpaca on open-source datasets, including from AI startup Anthropic, the fine-tuned stable LM models behave like chat GPT, responding to instructions sometimes with humor, such as write a cover

letter for a software developer or write lyrics for an epic rap battle song. The number of open-source text-generating models grows practically by the day, as companies' large and small vi-for visibility in the increasingly lucrative AI space. Over the past year, meta, invidia, and independent groups like the hugging face-backed big science project have released models roughly on par with private available through an API model such as GPT4 and Anthropics Claude, end quote. I mentioned today being 420 earlier in the show, and it looks like Elon's having a very 420 day himself. Potential war with Microsoft, as we mentioned, and today's Remember the

day, all the legacy blue checks go away unless you pay for them. And you might have seen the news that SpaceX had a rocket blow up this morning. It was a test of the most powerful rocket ever built, the starship rocket, the thing that's supposed to take humans to the moon and maybe Mars someday. This is the thing that SpaceX got the contract from NASA for. And yeah, it took off, but then, as I say, it blowed up. Or had rapid unscheduled disassembly, as SpaceX euphemistically said about it, which, as Marquez Brownlee tweeted, is sort of a friendly way of saying, it blew up, but we learned a lot on the way.

Indeed, this doesn't seem to be a disaster at all. SpaceX has been stressing the very experimental nature of this flight even before it happened, so, you know, boom, but apparently just getting off the launch pad was all they were really looking forward to as a step forward for this project. The video I saw had SpaceX people cheering even after the explosion, so I guess they'll have a trove of data to figure out

what didn't work for next time. Talk to you tomorrow.