☕️ Sam Altman is making moves

AI Tangle Newsletter

Good morning from the team at AI Tangle, hoping you have a great Tuesday! The AI world has served us quite a few developments over the weekend and early weekdays, with Zuckerberg eyeing artificial general intelligence development at Meta, politics taking a worrying approach thanks to some mischievous Silicon Valley entrepreneurs, and the birth of a new AI unicorn - ElevenLabs. Join us at AI Tangle as we untangle these intriguing early happenings.

OpenAI CEO Sam Altman plans to invest billions to build more AI chip factories​

Altman has been known to be a loud voice in accelerating semiconductor production for a long while now as AI becomes more pervasive, urging the industry to act immediately to ensure a steady supply of chips for continuous widespread developments near the end of the decade. To achieve this, the tech giant has recently been in talks about a global chip venture worth billions of dollars to set up large amounts of semiconductor fabrication plants.

​

Let's dive into the details.

Building and maintaining chip-fabrication plants, or fabs, isn't a popular nor favored approach among industry leaders like Google and Microsoft, who focus on designing custom silicon and then outsourcing the actual production of them to third-party suppliers. But as previously mentioned, Altman believes more is needed - current partners for the chip venture include Japanese investment firm SoftBank Group and Abu Dhabi's G42, the latter of which alone had focused on raising $8 billion to $10 billion, according to reports by Bloomberg. Potential partners also include Taiwanese TMSC and Samsung Electronics to help fabricate the chips, but all the parties mentioned have either declined to or have not been able to comment on the matter yet.

​Mark Zuckerberg pivots Meta to go big on AI​

Jumping aboard the artificial general intelligence (AGI) train next is Meta as CEO Mark Zuckerberg announced in an Instagram reel that the company will be joining Google's and OpenAI's alliance. To achieve this, Zuckerberg and Meta have begun the development of LLaMA 3, Meta's next-generation model, and additionally submitted an order for a ridiculous number of H100 Nvidia graphics cards, totaling 350,000 by the end of the year. However, much like with the case of Google and OpenAI, Zuckerberg does not have a clear definition for artificial general intelligence nor how to test whether they've achieved it, so it remains to be seen how this three-way tech giant race develops.

​

​DPD suspends AI chat feature after its bot goes rogue and swears at customer​

In a company statement reported by ITV, DPD said "The AI element was immediately disabled and is currently being updated" after its AI chatbot went rogue after a frustrated customer persuaded it to write a derogatory poem about the company's customer service. The aggravator, a pianist and conductor Ashley Beauchamp, shared his exchange with the bot on X/Twitter, where he told DPD's AI to tell him a joke after failing to get a status report on a parcel, later adding for it to write a derogatory poem about DPD's customer support and encouraging it to swear.

​

​Gmail might soon get an AI voice feature to simplify drafting emails​

Last year, Google introduced a new AI tool called "Help Me Write" for its Labs testing program in Gmail, proving largely successful. However, Google doesn't seem finished adding AI help just yet, as TheSpAndroid has spotted a potential new feature coming to the email service - drafting emails with your voice. Not to be confused with speech-to-text, which allows you to dictate the text being written, the new feature instead takes your voice as a prompt and drafts the email and all its content for you by giving you a dandy big microphone icon when you first start typing up a new email or replying to one.

​

​An alarming application of AI in politics as OpenAI bans developer of Dean.bot​

Recently, a ChatGPT-powered bot by the name of Dean.bot appeared, which took the liking of Rep. Congressman Dean Philips. Created by Silicon Valley entrepreneurs Matt Krisiloff and Jed Somers, who had started a super PAC (political action committee) called "We Deserve Better" to support Phillips ahead of the New Hampshire primary on Tuesday. However, despite disclaimers being set in place by the developers saying that Dean.bot was an AI tool and not Dean himself, researchers argued that it could still lure people in to accept a dangerous total, which ultimately caused OpenAI to strike down the ban hammer and enforce its usage policies, which forbids the use of the company's technology in political campaigns.

​

​ElevenLabs becomes the next AI unicorn after landing $80m in funding​

American-based AI voice cloning and software ​ElevenLabs​, created last year by ex-Google machine learning engineer Piotr Dabkowski and former Palantir deployment strategist Staniszewski, announced on Monday a monstrous Series B round of funding totaling $80m, backed by the likes of Andreessen Horowitz (a16z), former GitHub CEO Nat Friedman, and many more, putting the valuation of the startup now at $1bn. Known for its browser-based speech generation app that can create lifelike voices, ElevenLabs plans to increase detection measures regarding bad actors after some users on the infamous message board 4chan used ElevenLabs' tools to share hateful messages by mimicking celebrities.

​

​Elon Musk's AI company, xAI, received $500m in funding toward its $1bn goal... or did it?​

A recent Bloomberg report regarding a round of funding for xAI, an AI startup launched by Musk last July in response to excessive censorship, claims that the company has received $500m in funding towards its $1bn goal. According to Bloomberg, xAI is discussing evaluations of between $15bn and $20bn, though Elon and his group of investors are still a few weeks away from finalizing the terms, with some parties evaluating whether they can get computing power on top of, or just instead of, xAI equity shares. However, company founder Elon Musk says the whole report is outright wrong in an X/Twitter reply, ​leaving unanswered questions as xAI has yet to disclose the matter.

​DGM - A web app for the diagram enthusiast, DGM helps you create various diagrams with shapes and an AI-powered assistant, from initial empty sketches to fully-fledged formal notations.

​Assistive - Assistive is a platform for image and video generation - speak your mind or upload an image to let Assistive's AI create a video for you, be it a black and white cartoon or a hyper-realistic video clip.

​Antispace - Translating human thoughts to actions. Antispace is an action-oriented AI brainstorming partner, helping you dissect and understand problems, form solutions, and take charge of implementation.

​X-Me - X-Me is an entrepreneurial weapon that makes generating avatar videos easy - input some text, pick a voice, and create AI avatar videos hassle-free and in minutes,

A Killer Use Case (2-min read)

Blockchain, the technology behind cryptocurrencies like Bitcoin, could be vastly improved by implementing a blockchain-based AI training data ledger, finally giving the tech a "killer use case," says the CEO of Crypto Council for Innovation (CCI), Sheila Warren.

​

How Neural Networks Learned to Talk (27-min watch)

Art of The Problem, a YouTube channel focused on origin stories of modern ideas, dives into a 30-year retrospective of the language model and neural network scene. This video explores the journey of language models, from their modest beginnings through the development of OpenAI's terrifyingly powerful GPT models.