Google Strikes Back - H+ Weekly - Issue #414
This week - baby with DNA from three parents is born in the UK; EU is closer to passing AI Act; the first victims of ChatGPT; how cryopreservation can give you a new kidney or immortality; and more!
The first phase of the generative AI revolution belonged to OpenAI and Microsoft. Google was left behind and had to catch up.
But with what has been announced at Google I/O 2023 event, Google is back in the game.
Let’s start with Bard, Google’s chatbot and response to ChatGPT. Bard was in experimental mode since February 2023 and was only available via a waitlist. But now, Bard is available for free in 180 countries and territories. Surprisingly, Bard is not available for EU member countries. Google did not explain why is this the case but the best guess is that it can have something to do with the EU’s response to ChatGPT and the upcoming EU AI Act.
Bard soon will come with integration with tools from Google and third parties. It will be able to tap into Search, answer queries containing images and use Google Maps in responses. As an example of third-party integration, Google presented how Bard can tap into Adobe Firefly service and generate images from text.
Bard, and many other generative AI features Google announced, are powered by PaLM 2 - Google’s large language model. According to Google, PaLM 2 “excels at advanced reasoning tasks, including code and math, classification and question answering, translation and multilingual proficiency, and natural language generation”.
Generative AI features are also coming to Workspace. These new features include built-in text-to-image generators in Slides, and chatbot assistants in Docs and Sheets. Gmail will get the Help Me Write feature that can generate an email from simple text input. All those new features are in the experimental phase and available only through Google Workspace Labs.
Search is another place that got the generative AI makeover. Google is experimenting with including answers to the queries written by PaLM 2 in the search results with an option to ask follow-up questions and enter the conversational mode. These new features are available only for US residents via Search Labs.
Another interesting announcement was Project Tailwind, which was presented as an example of what is possible with PaLM API. It is an AI-powered notebook that takes in a document from Google Drive and creates a personal and private AI model to help better understand what is inside the document. Tailwind is available only to US residents and you have to join the waitlist to have a chance to check it out.
Google Cloud is also getting new generative AI services. Developers will soon be able to use PaLM 2 in their applications with PaLM API. Vertex AI, Google’s AI platform, is getting new models, including Imagen, Google’s text-to-image generator, and PaLM API for text and for chat. Other interesting models include Codey - a coding assistant, similar to GitHub Copilot, and Chirp - a speech-to-text model trained on over 300 languages.
Even Google Photos will get new generative AI features. With Magic Editor, you will be able to edit photos powered by generative AI.
There is also Magic Compose that can rewrite text messages to sound more cheerful or professional. Android is going to get a generative AI wallpaper generator.
An entire section of the event was dedicated to emphasising Google’s Responsible AI policy. To help fight misinformation, Google will be adding a new feature to Search that can give more information about an image, like when and where the image was posted. Google will be also adding metadata and watermarking AI-generated content to label them as such.
It will take a couple of months until everything presented at Google I/O 2023 will be available to the public. But Google is back and it brings state-of-the-art AI models and services deeply integrated within the Google apps ecosystem. The second half of 2023 in the generative AI space will be interesting.
From H+ Weekly
🦾 More than a human
Baby born from three people's DNA in UK first
A baby has been born using three people's DNA for the first time in the UK, the fertility regulator has confirmed. Most of their DNA comes from their two parents and around 0.1% from a third, donor woman. That 0.1% is there as an attempt to prevent the baby to be born with devastating mitochondrial disease.
German Doctors Are Attempting to Reverse Death and Resurrect Humans
Tomorrow Biostasis is the first cryopreservation company in Europe, which offers to freeze its customers’ bodies or brains and keep them in a cryopreservation state somewhere in Switzerland until the technology to revive them becomes available.
🧠 Artificial Intelligence
Stack Overflow is ChatGPT Casualty: Traffic Down 14% in March
Since the introduction of ChatGPT and coding assistants like GitHub’s Copilot, Stack Overflow, the biggest online forum for programmers to ask for help, has been seeing increasingly lower traffic. As SimilarWeb notes in this article, “on a year-over-year basis, traffic to Stack Overflow (stackoverflow.com) has been down by an average of 6% every month since January 2022 and was down 13.9% in March”. Meanwhile, GitHub CoPilot free trial signup page more than tripled from February to March.
AI race is disrupting education firms – and that is just the start
Education companies trading on the London and New York stock exchanges saw hundreds of millions wiped from their valuations after Chegg, a US firm that provides online help to students for writing and maths work, said ChatGPT was affecting customer growth.
EU nears ban on predictive policing and facial recognition after AI Act vote
This week, two committees of MEPs overwhelmingly endorsed sweeping new rules on artificial intelligence. The text now moves to a vote by the entire European Parliament in June. Once approved, the regulation will become the world’s first comprehensive AI law. The proposed AI Act now contains bans on biometric surveillance, emotion recognition, and predictive policing AI systems, as well as rules enforcing more transparency in AI systems, like disclosing that the content was generated by AI, designing the model to prevent it from generating illegal content and publishing summaries of copyrighted data used for training.
People are trying to claim real videos are deepfakes. The courts are not amused
Tesla is currently fighting in court a lawsuit brought by a family of a man killed when his Tesla crashed while using the self-driving feature. The family lawyers used Elon Musk’s bold statements about Tesla’s full self-driving feature against the carmaker. In response, Tesla’s lawyers claim that such statements, including a video of Musk boasting Tesla’s self-driving mode that was uploaded to YouTube six years ago, are deepfaked. The judge did not buy Musk’s lawyers' claims. "Their position is that because Mr. Musk is famous and might be more of a target for deep fakes, his public statements are immune," wrote the judge. "In other words, Mr. Musk, and others in his position, can simply say whatever they like in the public domain, then hide behind the potential for their recorded statements being a deep fake to avoid taking ownership of what they did actually say and do. The Court is unwilling to set such a precedent by condoning Tesla's approach here."
New Tech to Slash Organ Transplant Waiting Lists
In the US alone, 17 people die each day while waiting for organ transplants to save their lives. One of the biggest problems in organ transplantation is the short amount of time (measured in hours) the organ can be used after being removed from a donor. One startup got inspired by antifreeze proteins found in Arctic fishes and created a proprietary cryopreservative that can preserve human organs for much longer without the side effects of freezing, such as cells damaged by ice.
‘Remarkable’ AI tool designs mRNA vaccines that are more potent and stable
Scientists at the California division of Baidu Research have created an AI tool that optimises the gene sequences found in mRNA vaccines that could help to create jabs with greater potency and stability. In the validation tests, the new optimised vaccines triggered antibody responses up to 128 times greater than unoptimised vaccines. The algorithm also helped to extend the shelf stability of vaccines up to sixfold in standard test-tube assays performed at body temperature.
Can bioelectricity regrow limbs and organs?
Could bioelectricity be used to treat birth defects, traumatic injuries, limb loss, and perhaps even cancer? Michael Levin, a professor of biology at Tufts University and director of the Tufts Center for Regenerative and Developmental Biology, believes this is possible and has tested this innovative idea on flatworms and frogs.
▶️ How to catch a criminal cloner (1:14:13)
Here is the second part of the story of Hwang Woo-Suk - the scientist who convinced everyone to be a pioneer in human cloning and stem cell research, and who became South Korea’s hope for The Nobel Prize. The first part left us at the apex of Hwang Woo-Suk’s career, just as the lies and fraud started to be revealed. The second part tells the story of how Hwang Woo-Suk got exposed, how the South Korean government decided to put national pride over scientific integrity, the reaction of Korean society to the truth and what was the fallout of the biggest scientific scandal in South Korea’s history.
H+ Weekly is a free, weekly newsletter with the latest news and articles about AI, robotics, biotech and technologies that blur the line between humans and machines, delivered to your inbox every Friday.
Subscribe to H+ Weekly to support the newsletter under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).
H+ Weekly is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
A big thank you to my paid subscribers and to my Patreons: whmr, Florian, dux, Eric and Andrew. Thank you for the support!
Thank you for reading and see you next Friday!