Tech Brew Ride Home - Wed. 03/08 – The “Cambrian Explosion” Moment For AI

Episode Date: March 8, 2023

Today, you might have heard about the latest Elon controversy where he criticized an employee, and then walked it all back. We’ll get into that. DuckDuckGo melds two AI tools together. A Large Visual Language Model that will help robots move around in the real world. I explain why I’m telling you about every new Generative AI advance seemingly every single day. And with Google officially announcing the date for I/O, is their headlong rush to release AI products the waking of a sleeping giant, or the second coming of Google+? Sponsors: RelationshipHero.com/techmeme Links: FTC Twitter Investigation Sought Elon Musk’s Internal Communications, Journalist Names (WSJ) Twitter just let its privacy- and security-protecting Tor service expire (The Verge) Elon Musk apologizes after mocking disabled Twitter employee (AP) DuckDuckGo Releases Its Own ChatGPT-Powered Search Engine, DuckAssist (Gizmodo) Google’s PaLM-E is a generalist robot brain that takes commands (ArsTechnica) Google’s Plan to Catch ChatGPT Is to Stuff AI Into Everything (Bloomberg) 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 TechMeme right home for Wednesday, March 8th, 2023. I'm Brian McCullough today. You might have heard about the latest Elon controversy where he criticized an employee and then walked it all back. We'll get into all of that.

Duck. Dot. Go melds two AI tools together, a large visual language model that will help robots move around in the real world. I explain why I've been telling you about every new generative AI advance seemingly every single day. and with Google officially announcing the date for I.O. this year, is their headlong rush to release AI products, the waking of a sleeping giant, or the second coming of Google Plus? Here's what you miss today in the world of tech. Maybe if any one of these stories had popped up by its lonesome, I might have skipped doing the story individually. But since all three of these arrived all at once, let's just do a whole Twitter segment. Remember, in the background, to the whole saga is a consent decree that could literally throw Twitter into bankruptcy if it falls the

wrong way, if Twitter falls afoul of the consent decree. And according to the Wall Street Journal, the FTC has demanded Twitter turnover internal messages on Elon Musk, details about layoffs, and the names of journalists with access to company records, all in the name of that consent decree that's supposed to protect users. Quote, the FTC is also seeking to depose Mr. Musk in connection with the probe. We are concerned these staff reductions impact Twitter's ability to protect consumers' information, and FTC official wrote to Twitter's lawyers on November 10th, following an initial wave of layoffs, according to a copy of the letter viewed by the journal. The letters indicate Twitter responded to the FTC, but that the agency, as of late

January, felt the company was engaging in a troubling pattern of ongoing delay that raises serious concerns about its compliance, end quote. The FTC routinely seeks information that companies under a consent order provide third parties, including journalists, on grounds that the company couldn't withhold that same information from the FTC, Mr. Farrar said, end quote. Twitter has also let the certificate for its Tour Onion site expire, effectively killing off the privacy preserving and speech protecting service that it introduced just last year, quoting the Verge. Twitter no longer has a communications department to ask about the change, but the tour project confirmed the services lapse to the verge. The Onion site is no longer.

are available seemingly with no plans to renew. The Tour project has reached out to Twitter to look into bringing the Onion version of the social media platform back online, said Communications Director Pavel Zonoff in a statement. You can still visit Twitter.com via a browser running tour, but you won't get the added benefits. A tour-specific Onion site confers. Onion sites, sometimes called hidden services or dark websites, must be accessed via a browser that uses the anonymous and encrypted tour network. This keeps the user's web traffic and point of origin secret, and it also lets users get around government censorship efforts like those of Russia and China. Despite the Tor project's efforts to reach Twitter and resurrect the service,

its future doesn't seem rosy. The people who built it, at least all those I interacted with, are all gone, security engineer Alec Muffet, who helped launch the service last year, told the Verge over Twitter direct message, I'm pretty sure that it's going to stop working totally at some point unless Elon takes an interest, he said, end quote. And finally, you might have heard about this, but I'll let the AP take it from here. Quote, Haraldor Thorleafson, who until recently was employed at Twitter, logged into his computer last Sunday to do some work, only to find himself locked out along with 200 others.

He might have figured, as others before him have, in the chaotic months of layoffs and firing since Elon Musk took over the company, that he was out of a job. Instead, after nine days of no answer from Twitter as to whether or not he was still employed, Thorleafson decided to tweet at Musk to see if he could catch the billionaire's attention. and get an answer to his Schrodinger's job situation. Maybe if enough people retweet, you'll answer me here, he wrote on Monday. Eventually, he got his answer after a surreal Twitter exchange with Musk, who proceeded to quiz him about his work, question his disability and need for accommodations.

Thorlithson, who goes by the handle Holly, has muscular dystrophy and uses a wheelchair, and tweet that Thorleifson has a, quote, prominent active Twitter account and is wealthy. And the, quote, reason he confronted me in public was, to get a big payout, end quote. While the exchange was going on, Thorleafson said he received an email that he was no longer employed. Late Thursday afternoon, however, Musk had a change of heart. I would like to apologize to Holly for my misunderstanding of his situation. It was based on things I was told that were untrue or, in some cases true, but not meaningful, he tweeted. He is considering remaining at Twitter, end quote. Thorleafsson did not immediately respond to a message for comment following

Musk's tweet. In an earlier email, he called the experience surreal. You had every right to lay me off, but it would have been nice to let me know, he tweeted to Musk. Thorloveson, who lives in Iceland, has about 151,000 Twitter followers. Musk has over 130 million. He joined Twitter in 2021 when the company, under the prior management, acquired his startup UNO. He was lauded in Icelandic media for choosing to receive the purchase price in wages rather than a lump sum payout. That's because this way he would pay higher taxes to Iceland in support of its social services and safety net. Thorliefson's next move, quote, I'm opening a restaurant in downtown Reykjavik's very soon, he tweeted.

It's named after my mom, end quote. So the best sort of summary I can offer you is this. It seems like Elon knee-jerk assumed this guy was a slacker or a moocher or something. He tweeted, and this is Elon's tweet, quote. The reality is that this guy who is independently wealthy did no actual work, claimed as his excuse that he had a disability that prevented him from typing, yet was simultaneously tweeting up a storm. Can't say I have a lot of respect for that, end quote. To which, Holly responded, quote, I hope you are well. I'm fine too. I'm thankful for your interest in my health, but since you

mentioned it, I wanted to give you more info. I have muscular dystrophy. It has many effects on my body. Let me tell you what they are. My legs were the first to go. When I was 25 years old, I started using a wheelchair. It's been 20 years since that happened, and, in that time, the rest of my body has been failing me too. I need help to get in and out of bed and use the toilet. About nine years ago, I started a company called Unu.com. I worked a lot. It didn't do my body any favors, but it's what I felt I needed to do. The hard work paid off and the company became very successful. We worked for more or less every big tech company. We grew fast and made money. I think that's what you're referring to when you say independently wealthy.

That I independently made my money as opposed to, say, inherited an ever. Emerald Mine, end quote. Yes, he's apparently independently wealthy because Twitter bought his startup. Anywho, maybe because he took that buyout in salary, which would mean a big payout, I'm assuming if Elon fired him, or maybe the lawyers got to Elon and they were like, hey, it might not be legally kosher to fire a differently abled employee and then suggest that he wasn't doing any work or lying about his disability, or maybe someone just got to Elon and convinced him this was making him look like a jerk. So now we have Elon backtracking and making nice. And there we are. DuckDuck Go has launched OpenAI and Anthropic-powered Duck Assist into

beta, using Wikipedia, Encyclopedia Britannica and other sources to summarize answers. So it's interesting that they're pulling from two different LLMs here, combining them, according to Gizmodo. The tool is free and available on the Duck, Duck, Duck Go web browsing app for phones and computers, as well as the company's browser extension starting today. Unlike Microsoft's bungled AI projects with Bing, R-I-P-Sidney, Duck Assist isn't a chatbot. Instead, Duck Assist will suggest an automatic answer when it recognizes a search term it can answer. It's not being forced on anyone. When an AI-powered response is available, you'll see a Magic Wand icon with an Ask Me button

in your search results. The company says Duck Assist is still in beta, so it may not pop up that often yet. The company says limiting the data sources to Wikipedia will help stave off some of the bizarre misbehavior. Users have seen on other AI tools. For example, Microsoft had to limit the features of its new Bing Chat, which is also powered by Chat Chepti just over a week after its release. After reports showed Bing gave away its secret internal code name, users used racial slurs and gave out a long list of generally unhinged responses, Microsoft lobotomized the Bing AI and clawed back its functionality. privacy is the focus of DuckDuckGo's products, and the company says Duck Assist won't collect personal information, nor does it require a login, just like the company's other tools. Duck, DuckGo says it's keeping users search and browsing history anonymous in all data shared with OpenAI, the makers of ChatGPT and Anthropic, the company's partners on this project, end quote.

Meanwhile, Google and the Technical University of Berlin have unveiled Palm E, a visual language. model with 562 billion parameters integrating vision and language for robotic control. So this is a new sort of application of this stuff. Quoting Ars Technica, according to Google, when given a high-level command such as Bring Me the Rice Chips from the drawer, Palm E can generate a plan of action for a mobile robot platform with an arm developed by Google Robotics and execute the actions by itself. Palm E does this by analyzing data from the robot's camera without needing a pre-processed scene representation. This eliminates the need for a human to pre-process or annotate the data and allows for more autonomous robotic control. It's also resilient and can react to its environment.

For example, the Palm E model can guide a robot to get a chip bag from a kitchen and with Palm E integrated into the control loop, it becomes resistant to interruptions that might occur during the task. In a video example, a researcher grabs the chips from the robot and moves them, but the robot locates the chips and grabs them again. In another example, the same Palm E model autonomously controls a robot through tasks with complex sequences that previously required human guidance. Palm E is a next token predictor, and it's called PalmE because it's based on Google's existing large language model called Palm, which is similar to the technology behind ChatGPT. Google has made Palm embodied by adding sensory information. and robotic control. Since it's based on language models, Palm E takes continuous observations like images or sensor data and encodes them into a sequence of vectors that are the same size as language

tokens. This allows the model to understand the sensory information in the same way it processes language. Google Robotics isn't the only research group working on robotic control with neural networks. This particular work resembles Microsoft's recent chat GPT for robotics paper, which experimented with combining visual data and large language models for robotic control in a similar way. Robotics aside, Google researchers observed several interesting effects that apparently come from using a large language model as the core of Palm E. For one, it exhibits positive transfer, which means it can transfer the knowledge and skills it is learned from one task to another, resulting in, quote, significantly higher performance compared to single task robot models,

end quote. So having just told you that, gives me the opportunity to discuss something you might have been wondering about. Brian, why tell us about every new application, every new LLM that gets released every day, seemingly? Well, it's because I think there's a literal Cambrian explosion happening right now. Yesterday, all of these tools were, again, sort of parlor tricks. It was just about chatbots and cloning voices and making Dolly give you the picture of Einstein writing a penguin wearing S&M gear and the style of Renan Stimpy. But today, it's now about controlling robots. Tomorrow, who knows what it'll be? Like, either these AI systems are the new everything, meaning open AI or Anthropic are the new Google or the new meta, the new big tech

companies being born right now, taking their first steps right before our eyes. Or maybe they're the base infrastructure, the underlying tech stack that this new decade's worth of startups will be built on. So in that sense, I'm thinking of you developers and founders out there in the audience, trying to keep you abreast of the latest advances so you can be inspired to go off and start your own thing using these tools. But also, there's so many ways to think about this. Chris and I have this theory that we're calling varietals. Like, you know how in wine there are different grapes and flavors built on what valley a vineyard is in, what the soil is like, etc., etc.? Well, so will this be sort of like that? Somebody is going to build an AI tool that will be.

become the key tool for doctors to use to, you know, tell you what ails you. And whichever one wins, will it be because they trained their system on a particular set of medical research papers? Better than somebody else's system? Will you prefer to use some competitor to Dali because you like the style of art it produces more? But maybe Dali gets adopted in Hollywood because it's trained on different things. Varietals. Anyway, I'm keeping you abreast of all this stuff because it feels like history is happening right now in real time. I don't know who's going to win here or even what's going to win if we're in the first innings of this, or even if this is all just another hype cycle, if it's just a fad. But I'm trying to give you the lay of the land while it's happening so you can found and invest and apply for jobs accordingly.

Finally today, real quick, mark your calendars. Google has scheduled I.O. 2023 with a limited live audience for May 10th, when Android 14 details, Pixel 7A's debut, and a pixel tablet launch date are expected to be announced, and I would also say, expect the AI floodgates from Google to open here as well. In fact, Bloomberg has a look at Google's recent scramble to add generative AI to all of its products, which is reminding some employees of the rush to infuse Google Plus and social into every key Google service back around 2011. And we all know how that turned out. Read the whole piece, but, quote, The effort has CEO Sunder Pichai reliving his days as a product manager, as he's taken to weighing in directly on the details of product features, a task that would usually fall far below

his pay grade, according to one former employee. Google co-founders Larry Page and Sergey Bryn have also gotten more involved in the company than they've been in years, with Brin even submitting code changes to Bard, Google's chat GPT-esque chatbot. Senior management has declared a code red that comes with a directive that all of its most important products, those with more than a billion users, must incorporate generative AI within months, according to a person with knowledge of the matter. In an early example, the company announced in March that creators on its YouTube video platform would soon be able to use the technology to virtually swap outfits.

Some Google alumni have been reminded of the last time the company implemented an internal mandate to infuse every key product with a new idea. the effort beginning in 2011 to promote the ill-fated social network Google Plus. It's not a perfect comparison. Google was never seen as a leader in social networking, while its expertise in AI is undisputed. Still, there's a similar feeling. Employee bonuses were once hitched to Google Plus's success. Current and former employees say at least some Googler's ratings and reviews will likely be influenced by their ability to integrate generative AI into their work. The Code Red has already resulted in dozens of planned generative AI integrations.

We're throwing spaghetti at the wall, says one Google employee, but it's not even close to what's needed to transform the company and be competitive, end quote. In the end, the mobilization around Google Plus failed. The social network struggled to find traction with users, and Google ultimately said in 2018 that it would shudder the product for consumers. One former Google executive sees the flop as a cautionary tale, quote, the mandate from Larry was that every product has to have a social component. this person says. It ended quite poorly, end quote. But in the long run, it may not matter much that Open AI sucked all the air out of the public conversation for a few months, given how much work Google has already done. Pachai began referring to Google as an AI-first company in 2016. It's used machine learning to drive its ad business for years, while also weaving AI into key consumer products such as Gmail and Google Photos, where it uses the technology to help users compose emails and

organize images. In a recent analysis, research company Zeta Alpha examined the top 100 most cited AI research papers from 2020 to 2022 and found that Google dominated the field. The way it has ended up appearing is that Google was kind of the sleeping giant who is behind and playing catch-up now. I think the reality is actually not quite that, said Amin Amad, a former AI researcher at Google who co-founded Vectara, a startup that offers conversational search tools to businesses. Google was actually very good, I think, at applying this technology into some of their core products years and years ahead of the rest of the industry, he said, end quote. Nothing for you today. Talk to you tomorrow.