Five Minute Blockchain – No. 56
01.06.2023
Estimated reading time: 6 min 12 sec
Quote of the Week
“Mitigating the risk of extinction from A.I. should be a global priority alongside other societal-scale risks, such as pandemics and nuclear war.”
One-sentence statement signed by AI experts, published on the Center for AI Safety webpage. LINK
TRUST
AI experts warn about AI
This week’s big trust/content/tech topic is puzzling: Why would experts warn about the potential threats of AI? After all, this is the technology they are working on, in a race one against the other to make the technology better.
Essentially, founders like Sam Altman urge that AI platforms should be regulated. Another idea is that AI platforms should be required to have a license to operate from a government agency which has yet to be established.
The New York Times ($)
BBC
Ars Technica
CONTENT
Google rolls out AI/LLM powered “Search Generative Experience” to some users.
As announced during the recent Google Event I/O 2023, users who have signed up for the Search Labs program can now use “Search Generative Experience” (SGE) – a Google search supported by AI.
The key idea of SGE is to support users with “longer, multistep searches. These queries might not necessarily be answered by one website.
Below is an illustrative example of how this could work:
“Maybe you’re starting to map out a decision that you’d typically need to break down into smaller parts, like ‘Learning ukulele vs guitar.’ Search will provide an AI-powered snapshot so you can get help understanding what factors to consider.”
9to5 Google
BabyLM Challenge aims to create powerful AI with small language models
ETH Zurich and others are part of a particular challenge to perform research towards small but effective language models for AI. This is interesting because large current models can only operate with extensive cloud resources. Smaller models could be easier to handle while still providing lots of output.
The New York Times ($)
BabyLM Challenge (Github)
How Adobe adds “product” to AI
The big tech and software companies are racing to react to generative AI. Last week Adobe introduced its offering, called Firefly. The software enables the generation of images and is part of the well-established Photoshop. Firefly enables “generative fill”, providing many new options for visuals and pictures.
Benedict Evans on how this is different from other offerings.
“Adobe made a very successful shift to subscription SaaS in the last decade. Now it’s trying the same with generative AI, launching a de novo image generation product in Firefly and adding generative features to Photoshop.
The more generally important part of this, I think, is the move to add interface, control and product to the prompt: instead of typing 50 words into a box and waiting to see what you get, there are options and switches to give you some control.”
Adobe Firefly
Examples of how Firefly can be used (Thread on Twitter)
Was Google Bard trained on Gmail data? Yes, no, maybe?
Asked what data was used for training, Bard, the AI software from Google, recently replied to a user that Gmail data was used. This might have been a case of AI hallucination, where platforms make up answers. But even the potential use of (private) Gmail data set off a flurry of comments and articles, as many users and journalists would consider this a breach of privacy … and trust. Google wanted to ensure that Bard was not trained in using Gmail data and that the software was “hallucinating” in this particular answer. But the honest answer is a bit more complicated. A good write-up of the whole situation can be found under the link.
Skiff
Apple App Store generated $1.1 trillion of commerce in 2022
Apple released a study with data about the amount and type of revenue generated globally in the App Store. To this end, the company commissioned a research firm for a report, which can be downloaded for free.
The numbers are very, very high. $1.1 trillion (Europe, long ladder = one billion)
“Apple says its App Store ecosystem generated $1.1 trillion in developer billings and sales in 2022, 90% of which was commission-free — a metric it likes to tout to downplay the growing complaints about the high cost of doing business on a marketplace that generally takes a 15% to 30% commission on in-app purchases and paid downloads, with some exceptions.
This $1.1 trillion breaks down as $910 billion in total billings and sales from the sale of physical goods and services, $109 billion from in-app advertising and $104 billion for digital goods and services.”
In 2019 the full-year revenue was “only” $519 billion (Europe: Milliard), which means the figure doubled during the pandemic.
TechCrunch
Apple Press Release
Analysis Group (PDF)
An app for short attention spans
“Web Roulette, a mobile web browser app for iOS built for the short attention spans of the TikTok era. With the debut version out now, you can add your favorite websites or choose from its suggestions, then swipe through the sites to see what’s new or shake the app for a surprise web page when boredom strikes.”
TechCrunch
Digital nomads are a new type of tourist but with some adverse effects on local communities
Cities from Canggu to Medellín are welcoming tech workers, but locals complain they’re being priced out.
Rest of World
BLOCKCHAIN
Fireside Chat: A Twitter competitor with (potentially) 45 million users
A key element and difference to Twitter are rewards for good entries:
“Created by Pi Network and dubbed Fireside Forum, the initially text-based social platform is enlisting a model where users can spend tokens rewarding and elevating posts they like or invest them in penalizing the posts they don’t. The company’s aim is the Pi Network model will help suppress content considered spam, misleading, toxic or all of the above.”
The Block
White Paper from World Economic Forum suggests ways to regulate Crypto
Here is a new approach to creating trust and some form of regulation for crypto assets:
“The World Economic Forum (WEF) recently published a white paper titled “Pathways to the Regulation of Crypto-Assets: A Global Approach,” advocating for a collaborative approach towards crypto regulation on a global scale. The white paper highlights the unique challenges and necessary considerations regarding the regulation of crypto-assets. Considering the borderless, open-source, decentralized nature of these digital currencies, their regulation requires a delicate balance between preventing harm, protecting users, and promoting innovation.”
World Economic Forum – Pathways to the Regulation of Cryptoe-Assets: White Paper
Tokenization gains popularity in finance, elsewhere
“Asset tokenization is increasingly being tested to improve costly capital markets operations like securities issuance. New York-based Fireblocks, which helps businesses create and manage digital assets, believes tokenization will enjoy exponential adoption over the next two years, as early-adopter businesses start to experience the very tangible benefits.”
Blockdata
SAP to start NFT platform
“SAP sees NFTs as an extension of digital twins where the twin is a promo item, a ticket for an event, or can be redeemed for physical goods.”
An initial trial has been done with a German retailer in the area of customer loyalty:
One of the companies that SAP has worked with to build the solution is Tchibo, the German coffee shop and online retailer with more than 1,000 stores and revenues of more than €3 billion.
… As a trial, Tchibo minted 1,000 NFTs on the Polygon blockchain for its Royalty Club. Some Golden NFTs were redeemable for small prizes.
Ledger Insight
Short Links
Thank you for reading. If you have questions or suggestions, please contact us via info@trublo.eu.
Photo by Daiga Ellaby on Unsplash
Five Minute Blockchain – No. 55
16.05.2023
Estimated reading time: 7 min 55 seconds
QUOTE OF THE WEEK:
TRUST
Transparency and safety: European Parliament preparing a set of rules for AI and surveillance technology
In the past, rulings and regulations for new technologies lagged, sometimes for years. But with Artificial Intelligence (AI), things seem to move faster. Last week two committees of the European Parliament adopted a draft of far-ranging rules for the use of AI and surveillance technology.
Once approved, these could become the “world’s first rules on Artificial Intelligence”. The rules include the right to complain about AI systems and provide a handle for system changes. Another element is the proposed ban on “biometric surveillance, emotion recognition, predictive policing AI systems.”
In a press release, European Parliament said the goal is to ensure a human-centric and ethical development of AI in Europe.
“MEPs aim to ensure that AI systems are overseen by people, are safe, transparent, traceable, non-discriminatory, and environmentally friendly.”
“AI systems with an unacceptable level of risk to people’s safety would be strictly prohibited, including systems that deploy subliminal or purposefully manipulative techniques, exploit people’s vulnerabilities or are used for social scoring (classifying people based on their social behaviour, socio-economic status, personal characteristics).”
Below is a list of technologies and practices which would be banned under the future EU rules:
- “Real-time” remote biometric identification systems in publicly accessible spaces;
- “Post” remote biometric identification systems, with the only exception of law enforcement for the prosecution of serious crimes and only after judicial authorization;
- Biometric categorisation systems using sensitive characteristics (e.g. gender, race, ethnicity, citizenship status, religion, political orientation);
- Predictive policing systems (based on profiling, location or past criminal behaviour);
- Emotion recognition systems in law enforcement, border management, workplace, and educational institutions; and
- Indiscriminate scraping of biometric data from social media or CCTV footage to create facial recognition databases (violating human rights and right to privacy).
European Parliament Press Release
TechCrunch
Google deploys pass keys
Google is extending the roll-out of pass keys to all Google accounts. This is part of a broader move away from hackable/often insecure “12345” passwords, with similar pushes towards more security expected from Apple, Microsoft and others.
From a Google Press Release:
“Passkeys are a new way to sign in to apps and websites. They’re both easier to use and more secure than passwords, so users no longer need to rely on the names of pets, birthdays or the infamous “password123.” Instead, passkeys let users sign in to apps and sites the same way they unlock their devices: with a fingerprint, a face scan or a screen lock PIN. And, unlike passwords, passkeys are resistant to online attacks like phishing, making them more secure than things like SMS one-time codes.”
Google
Conference: Future of AI
On 29 June 2023, Horizon Europe research projects AI4media, AI4Trust, TITAN and vera.ai – in cooperation with the European Commission – host a one-day event focusing on various facets of Artificial Intelligence and the disinformation landscape. Full title: Meet the Future of AI: Countering Sophisticated & Advanced Disinformation.
Registration
CONTENT
Media company Vice files for bankruptcy
The media start-up was once valued at $5,7bn but had recently seen less revenue from digital advertising. The company websites will keep operating until a buyer is found.
“Investments from media titans like Disney and shrewd financial investors like TPG, which spent hundreds of millions of dollars, will be rendered worthless by the bankruptcy, cementing Vice’s status among the most notable bad bets in the media industry.”
The New York Times ($)
Employees at Microsoft like the new training videos as much as a Netflix series
Training videos are usually dull – you must watch them, but it is not easy. Someone or a team at Microsoft has found a better way: Training videos that are so interesting and well-done that people like to see them.
The series is called “Trust Code” and is now in its 7th season. The main character is played by an aspiring actor named Devin Badoo- a star, at least among many of the 220.000 Microsoft workers:
“For employees at most companies, sitting through training videos every year is about as welcome as a toothache. “Trust Code,” with its recurring characters and end-of-season cliffhangers, is redefining the genre. Since launching in 2017, it has inspired watch parties, viral memes and T-shirts with Mr. Badoo’s image.”
The Wall Street Journal
AI Update
Last week Google had its big annual developer conference and launched several significant updates for AI technology in the search platform.
Here are three links to get you a quick update here:
Google shows AI features coming to Search, including an AI-powered “snapshot” that summarizes search results with links to sites “corroborating” the information
Google makes Bard available in English in 180 countries and territories, promises AI image generation from Adobe and integration with services like Instacart
Google launched a dedicated Labs page where users can sign up to test Google’s early ideas for features and products, including Search and Workspace AI tools.
Nearly half of YouTube views in the US are on TVs
YouTube is growing. There is a new generation of content creators, resulting in a constant flow of exciting video content for almost any interest and niche. The audience has noticed YouTube is on its way as an even bigger competitor to traditional TV.
“Internal data indicate that close to 45% of overall YouTube viewing in the U.S. today is happening on TV screens”.
The Information
BLOCKCHAIN
Blockchain Large Language Models
How to use blockchain as a tool for intrusion detection:
“This paper presents a dynamic, real-time approach to detecting anomalous blockchain transactions. The proposed tool, BlockGPT, generates tracing representations of blockchain activity and trains from scratch a large language model to act as a real-time Intrusion Detection System.”
Arxiv
The Crypto Trash Moat
A big question: To what extent are crypto platforms used for crime?
There can only be estimates: “Any conversation about crypto and crime needs to disclose that, according to the folks with the data, less than 1% of total crypto transactions can be tied to illicit use – at least that’s what Chainalysis reports.”
What is going on a lot in crypto and elsewhere are “confidence games”, where criminals lure people into paying for something (often as an investment) and then trick them out of their money.
In 2022 the US Department of Justice named cybersecurity expert Eun Young Choi as the first director of the “National Cryptocurrency Enforcement Team (NCET)”. The department seems to focus on “smaller issues” like social media scams, fraudsters and darknet misuse”.
Instead of major headline-drawing scandals like FTX and 3AC, Choi’s department seems primarily focused on relatively smaller issues like social media scammers, darknet misuse and online fraudsters – an activity that’s rarely discussed openly but which exists as a sort of background hum for anyone spending time on Crypto Twitter and Discord. (Paul Dylan-Ennis, a frequent contributor to CoinDesk, calls this crypto’s “trash moat …)”
The amounts of money the NCET and other US departments are securing are substantial:
While scams like these often only damage a single victim at a time, it can still be big money. NCET, along with other agencies, booked upwards of $112,000,000 from busting six such U.S.-based scams. The Federal Bureau of Investigation (FBI) estimates $3.31 billion was stolen from people in 2022 through investment fraud, with crypto-related scams accounting for more than a third (~$2.57 billion) of that figure. Worse than just money lost, the proliferation of confidence games – which require bad actors to cultivate long-term relationships and build trust with their marks – has tainted crypto’s reputation.
Coindesk
Thank you for reading. If you have questions or suggestions, please get in touch with us via info@trublo.eu.
Five Minute Blockchain – No. 53
23.03.2023
Estimated reading time: 7 min 14 sec
A weekly update from TruBlo –> https://www.trublo.eu
QUOTES OF THE WEEK:
Too much content about Web3 platforms, tools and stats. Not enough about artists.
@musicben_eth (Twitter User)
“We’re at a tipping point where AI is going to break trust in what you see and hear — anddemocracies can’t survive when people don’t agree on facts. You have to have a baseline of understanding of facts”
Dana Rao, Adobe’s general counsel and chief trust officer (TechCrunch)
TRUST
Policy update for Open AI Codex will affect research papers
Scientists need access to machine learning (ML) models to conduct research. One of the access points is Codex, by OpenAI, which is not open source.
On Monday, OpenAI announced that it would discontinue support for Codex by Thursday. Hundreds of academic papers would no longer be reproducible: independent researchers would not be able to assess their validity and build on their results.
AI Snake Oil
Who is Shou Zi Chew, the TikTok CEO?
This Thursday, the boss of TikTok will be at a hearing in the US, where politicians want to decide about a potential ban on TikTok, mainly based on security concerns.
The 40-year-old CEO has worked at Goldman Sachs, studied in the US and manages TikTop from an office in Singapur. The “Wall Street Journal” has a profile of the executive.
TikTok argues that the company would never share user data with Chinese authorities, and there is little trust in this claim in the US, Europe, and India.
Wall Street Journal
TechCrunch
Blacklight – a real-time website privacy inspector
On the pages of The Markup, you can access a tool to check the privacy settings on any website, including your own. Service is free.
The Markup
The new Luddites
Between 1811 and 1816, English workers repeatedly destroyed new machines, specifically in cotton and woollen mills, because the new technology threatened their jobs. They were called “Luddites“.
Now there is a new group, not as violent as their predecessors, but with a similar opinion about new technologies. In New York, a group has formed the “The Luddite Club”. They promote “self-liberation from social media and technology”. Part of that is to evade smartphones by using flip phones from the 90s, meeting in person, and to read from physical books.
According to a survey “68.6% of people say that screen time has had a negative effect on their mental health, while one in three admitted it had a detrimental effect on their work or personal lives. In response to this, we’re seeing the emergence of a new movement of people, groups and brands who are questioning technology’s unchecked impact on our lives, with many championing a return to tactile, physical, off-line experiences.”
Wunderman Thompson: The New Luddites
CONTENT
You can’t make this up
The new AI platforms are the big topic in early 2023. How can the new options be used? What are the shortcomings? What are the dangers to jobs, society and for truth and trust?
Here is a quick run-down of exciting headlines this week:
- RightWingGPT is an AI model trained on conservative political viewpoints. Its creator thinks that ChatGPT has a “left-leaning bias. (New York Times, $)
- Canva, the design platform, introduces AI-based tools to simplify creating images or presentations by writing prompts. The Australian company, founded in 2012, has around 125 million users, with revenue of $1,6B annually. (Forbes).
- GitHub has announced Copilot X, which means AI support for developers. OpenAI GPT-4 is used to facilitate coding tasks. (Bloomberg)
- News Publishers want their share. An alliance of media companies are discussing how to demand a percentage of the revenue from Microsoft, Google and other AI platforms for using their content to train the AI software. (Wall Street Journal)
- Character.AI faces legal troubles for allowing users to make and share chatbots of famous people or fictional movies, or comic characters such as Yoda or Harry Potter. (Bloomberg)
- Ubisoft has launched an AI tool called Ghostwriter to assist video game developers with writing in-game dialogue. (Kotaku)
- First Batch ID (FID) is a decentralised identity to personalise a user’s news feed. (Gravitates)
Google and Microsoft plan to enhance office software with AI magic
AI software might change how work is done for millions of jobs by offering new options for typical office tasks, such as writing an e-mail or a marketing message. Both Google and Microsoft are in a race to deploy such features:
In a replay of last month’s dueling chatbot launches by the tech giants, Alphabet touted a “magic wand” for its popular Google Docs software that can draft a marketing blog, training plan or other text, then revise its tone at users’ discretion, a company official demonstrated to reporters.
Alphabet announced that the AI software could summarise messages in Gmail, craft slide presentations, personalise customer outreach and take meeting notes.
Meanwhile, Microsoft organised an event to discuss how it is “reinventing productivity with AI”. Like Google, the company plans to connect AI options to Word, Powerpoint and Excel.
It can be expected that there will be a debate about the amount of AI-generated content for a while. Still, with such deep integration into everyday life, most people might get accustomed to using AI to speed up the completion of standard office tasks.
Reuters /via Yahoo
Glaze aims to prevent AI crawlers from stealing art
A free tool called Glaze, developed by a team from the University of Chicago, seeks to help artists to prevent AI from copying their art.
“…the (beta) app works by adding almost imperceptible ‘perturbations’ to each artwork it’s applied to — changes that are designed to interfere with AI models’ ability to read data on artistic style — and make it harder for generative AI technology to mimic the style of the artwork and its artist. Instead systems are tricked into outputting other public styles far removed from the original artwork”.
Glaze Download
TechCrunch
BLOCKCHAIN
Blockchain technology used for refugee aid programs in Ukraine
The UN refugee program is among the pioneers of using blockchain technology for aid programs, specifically for transferring financial aid to refugees. The agency had already used similar technology to support refugees in Jordan in 2018, but with a different setup.
In Ukraine, the UNHCR works with the Stellar Development Foundation, a nonprofit alongside the money transfer company MoneyGram. A fourth organisation involved is Circle Internet Financial, the issuer of USDC stablecoin.
“For now, the stablecoin program in Ukraine is being piloted on a microscopic scale, with fewer than 100 participants in the cities of Kyiv, Lviv, and Vinnytsia. The UNHCR is preparing to expand the initiative to up to 5,000 wallets by April, but this would still represent only a fraction of the number of Ukrainians displaced by the war.”
Distributed technology and digital wallets can solve several challenges, such as handling and distributing large amounts of cash.
UNHCR Press Release (December 2022)
Wired ($)
US authorities send notice to Coinbase
Coinbase and Tron received SEC notices that the exchanges might have violated US securities laws by selling securities. Coinbase said that the warning does not mean any changes to current offerings. (CNBC)
“Brian Armstrong, the co-founder and CEO of crypto exchange Coinbase, has compared the United States Securities and Exchange Commission (SEC) to “soccer refs” in a game of pickleball, criticizing U.S. regulators for not being able to “agree on the rules” of “this new game.”” (Cointelegraph)
SEC sues Tron-founder and eight celebrities
Meanwhile, the SEC sued Justin Sun, the founder of Tron, for selling unregistered securities, market manipulation, and fraud. Also, eight celebrities are charged for promoting crypto assets without disclosing that they were paid for it. (SEC Press Release)(Coindesk)
Luxembourg recognises bill for DLT-based securities
Last week the Luxembourg parliament adopted a bill that recognizes DLT-based securities as collateral. It also expands the definition of financial instruments to include those issued under the EU’s DLT Pilot Regime, which comes into force later this month.
Ledger Insights
SHORT LINKS
- “A Moody’s report said municipalities could save money and improve transparency using blockchain technology. The firm added that doing so comes with potential risks, namely cyber attacks and volatility.” (The Block)
- “South Korea cranks up pressure to extradite Do Kwon, founder of failed Terra-Luna stablecoin”. (Forkast)
- In Nigeria: “A youth-led online movement propelled an outsider candidate into the political mainstream.” (Wired UK)
- On the launch of Microsoft Loop, which is partially similar to Notion: “If you create software that gains traction in work environments, it’s inevitable that Microsoft is going to follow.” (Daring Fireball) (The Verge)
- Status of Neobanks in Europe -profits, outlook (Tech.eu)
- Flying Sheep Studios announced that the Cologne-based studio had received $1.2 million from a German ministry for “Star Life”, a social massively multiplayer online (MMO) game. (Venture Beat)
- Chainalysis breaks down how scammers adapt during the bear market (Cointelegraph)
- Ponzi 2.0: “An investment fund announces a $16 million investment in an Estonian startup. But there’s a problem — the startup doesn’t exist, and the funding didn’t happen.” (Tech.eu)
Thank you for reading. If you have questions or suggestions, please contact us via info@trublo.eu.
Photo by Mahdis Mousavi on Unsplash
Five Minute Blockchain – No. 52
The weekly newsletter from the crossroads of trust, content & blockchain.
TruBlo is an EU research project. We support 45 early-stage projects to research and create new solutions.
13.03.2023
Estimated reading time: 9 minutes 44 seconds
QUOTE OF THE WEEK:
“A platform that can mimic humans’ writing with no commitment to the truth is a gift for those who benefit from disinformation. We need to regulate its use now.”
Emily Bell: A fake news frenzy: Why ChatGPT could be disastrous for truth in journalism
The Guardian
TRUST
Bank failures in the US affect start-ups and crypto platforms
In the US, three banks defaulted in the past week. First Silvergate, then Silicon Valley Bank, and then – only since Friday – another bank called Signature was closed in New York. The reasons are different, though three have in common that creditors lost trust. It is bad enough for the US government to step in.
After receiving a recommendation from the boards of the Federal Deposit Insurance Corporation (FDIC) and the Federal Reserve, Treasury Secretary Yellen, after consultation with the President, approved actions to enable the FDIC to complete its resolutions of Silicon Valley Bank and Signature Bank in a manner that fully protects all depositors, both insured and uninsured. (Board of Governors of the Federal Reserve System)
The downfall of Silicon Valley Bank will have repercussions for venture capital and start-up financing, potentially around the world. If the bank behind many deals in this space suddenly fails, there is a need to investigate the causes.
The unraveling at Silicon Valley Bank will have far-reaching implications for U.S. venture-backed startups, half of which did business at the bank, and the broader tech ecosystem. (Crunchbase)
The lightning collapse of Silicon Valley Bank Friday raised the specter of a broad tech-industry crash for the first time since the dotcom bubble burst in 2000. That threat, which loomed all weekend as legions of the startups that made up the bank’s clientele worried about meeting next week’s payrolls, receded after the federal government intervened Sunday to backstop depositors’ assets even over the $250,000 FDIC threshold. After a year of layoffs and market retreats, the run on the industry’s own community bank put tech’s new status as a troubled business in sharp relief. (Axios)
In the UK, HSBC bought the British subsidiary of Silicon Valley Bank for one British pound (New York Times, €)
Crypto platforms are also involved because some might have exposure to these failing banks. Circle and Coinbase published statements to assure clients can redeem their holdings this Monday.
The shutdown of Signature and the collapse of Silvergate leave many companies in the crypto industry without much access to the US banking system. (Web3 is going just great)
Board of Governors of the Federal Reserve System
Joint Statement by the Department of Treasury, Federal Reserve and FDIC
Crunchbase
Axios
The Economic Times
CNN
The Verge
Web3 is going just great
Fractional ownership investments are popular in South Korea
After each financial crisis (2003, 2008), one would think that there is a lesson learned about making a lot of money, fast. But there are repeating patterns, and of course, there is pressure on the next generation to make enough money somehow, usually to buy a flat or a house.
So, in South Korea, there is an early cycle of investing, which – as of now – is not very critical but could be the nucleus of another crisis and loss of trust in five to ten years:
“Young South Koreans are unusually keen to put their meagre savings to work—perhaps because the country’s lacklustre jobs market makes it hard for them to buy property or support a family. A survey in 2021 found that eight in ten people in their 20s and 30s invest in stocks, cryptocurrencies or other assets. And they start young; 7% of shareholders in Samsung Electronics, a tech manufacturer, are in their teens.”
The Economist (€)
CONTENT
Meta is developing a text-based decentralised social network as a Twitter competitor.
Meta is working on a text-based social network. Based on current information, the platform will use ActivityPub, a decentralised social networking protocol which powers Mastodon and other decentralised apps. The codename, for now, is P92, and the platform will be Instagram-branded. The current plan is to publish a minimum viable product with no clear release date.
Cointelegraph
The week in generative AI
Generative AI is so quickly evolving right now that it is challenging to keep track. However, here is a quick update on some of the most relevant stories this past week:
- ChatGPT and Whiper are accessible via an API (OpenAI) at meager prices. (Nathan Labenz)
- Antrophic, another AI company founded by ex-Open AI employees and funded with $124 million, is offering its language model to other companies. (TechCrunch)
- Elon Musk is considering setting up a rival, despite being one of the co-founders of OpenAI. (The Information)
- Generative AI helps developers to boost productivity (WSJ) (€)
- Video: How Nvidia powers the AI revolution (CNBC)
- A developer manages to run the LLaMA model on his laptop. LLaMA was developed by Meta for $13B and is considered competitive to GPT-3 from OpenAI. Costly hardware is needed to run GPT-3. (Simon Willison)
- Using LLaMA with M1 Mac (l1x/dev)
- Dalai is a “dead simple way to run LLaMA on your computer (Cocktailpeanut.Github.io)
- Neal Mohan, the new CEO of YouTube, released a statement saying that YouTube will introduce new AI-driven features for creators (YouTube Blog)
Come on, Google, let’s dance: Bing says it crossed 100 DAUs
Since introducing Chat GPT to assist with search queries four weeks ago, Bing has experienced user growth. According to Microsoft, the search engine crossed the number of 100M DAUs (Daily Active Users), though the blog post did not specify the number before introducing new AI features.
“Of the millions of active users of the new Bing preview, it’s great to see that roughly one third are new to Bing. We see this appeal of the new Bing as a validation of our view that search is due for a reinvention and of the unique value proposition of combining Search + Answers + Chat + Creation in one experience.”
For comparison: Google is still the market leader, with a market share in many countries in the 95% and above range. But Bing had already gained some market share in the past and will grab more. We are witnessing a new phase of the competition, not only for search but for the quality of answers and information retrieval. New generative AI is just the opening for intense competition in the search business. There is one particular sentence in the blog post from Microsoft clearly showing how this has energised the company:
This (the number of DAUs, sic) is a surprisingly notable figure, and yet we are fully aware we remain a small, low, single digit share player. That said, it feels good to be at the dance!
Bing Blog
Grammarly announces GrammarlyGO which uses generative AI to write & rewrite content
Launching in the coming months, “GrammarlyGO” is a generative AI product that is designed to “accelerate productivity where people write.”
Like ChatGPT, GrammarlyGO is able to create text based on a short prompt, though Grammarly’s special trick is that the content generated copies your usual writing style – after all, Grammarly already analyzes everything you write for typos, so there’s plenty of data to work with. Use cases for this that Grammarly points out includes writing email replies based on one-click prompts such as “I’m not interested.”
9to5 Mac
The “Bold Glamour” TikTok Filter is highly problematic
A newly released TikTok Filter called “Bold Glamour” will change your appearance from ordinary to model features. The generated visual looks highly realistic and works dynamically in a short video, and the superficial perfection draws considerable criticism. It is one thing to be confronted with beautiful people on magazine covers, even when you know all images are heavily worked on in Photoshop. But it is another escalation when new filters can make you a “beauty” with one click. According to at least one study, comparing machine-generated images and actual appearance leads to a desire for plastic surgery.
CNN
When influencer parents use their kids to produce content
Quote:
Claire, whose name has been changed to protect her privacy, has never known a life that doesn’t include a camera being pointed in her direction. The first time she went viral, she was a toddler. When the family’s channel started to rake in the views, Claire says both her parents left their jobs because the revenue from the YouTube channel was enough to support the family and to land them a nicer house and new car. “That’s not fair that I have to support everyone,” she said. “I try not to be resentful but I kind of [am].” Once, she told her dad she didn’t want to do YouTube videos anymore and he told her they would have to move out of their house and her parents would have to go back to work, leaving no money for “nice things.”
When the family is together, the YouTube channel is what they talk about. Claire says her father has told her he may be her father, but he’s also her boss. “It’s a lot of pressure,” she said. When Claire turns 18 and can move out on her own, she’s considering going no-contact with her parents.
Teen Vogue
YouTube and the multi-million-dollar dubbing economy
From “Rest of World”:
- Unilingo, the dubbing provider for MrBeast and PewDiePie, is part of a new localisation strategy for some of YouTube’s biggest stars.
- YouTube dubbing is helping creators reach new audiences worldwide and monetise the same video in several languages.
Rest of World
BLOCKCHAIN
European politicians suggest building a European blockchain called “Europeum.”
Mathieu Michel, Belgian Minister digital Minister, talked to Coindesk, offering a blockchain solution for infrastructure and public services:
“After regulating cryptocurrencies, European politicians are contemplating the next step in the race to attract Web 3 business – and it might be a tailor-made blockchain that respects privacy, Belgium’s digital minister told CoinDesk in an exclusive interview. A new “Europeum” blockchain could be the vehicle to record property ownership, driving licenses or professional qualifications while sticking to the European Union’s high regulatory norms, Mathieu Michel said.”
Coindesk
USDC stablecoin depends on Silicon Valley Bank collapse
The financial services firm Circle confirmed Friday that $3.3 billion of its reserves were tied up at Silicon Valley Bank, resulting in the USDC stablecoin going below one US dollar.
“Stablecoins derive their value from those reserves; if one is worth more than $43 billion – as USDC was earlier on Friday – there should be roughly that much cash or cash-like fixed-income instruments stashed somewhere backing that up. USDC’s market capitalization has now slumped below $40 billion.”
Crypto was born in the aftermath of – and, to some, in response to – the 2008 crisis. Satoshi Nakamoto’s Bitcoin paper debuted into a world where governments had just propped up the financial system by pouring money into it. Crypto lacks such a centralized authority. If SVB customers, including Circle and its USDC stablecoin, are forced to take a haircut on their money, the repercussions are unclear.
Coindesk
Crypto.com lost the ability to accept US-Dollar.
Crypto.com announced it could only provide euro-denominated banking last week. The exchange lost its ability to accept US-Dollar deposits due to problems with banking partners. Gemini denied that the relationship with JPMorgan had ended in a tweet.
Cointelegraph
SHORT LINKS
- PeopleDAO loses $120,000 after the payment spreadsheet is shared publicly (Web 2 is going just great)
- Meta to end Canadians’ access to news on Facebook and Instagram if Bill C-18 becomes law (The Globe and Mail)
- WhatsApp has started a fight with the UK about encryption (Wired UK)
- A conservative Catholic group in Colorado bought a mobile app tracking data worth millions of dollars to identify gay priests across the US. (PC Mag)
- More than €1.4 billion was invested in European Tech this week (Tech. eu)
Thank you for reading. If you have questions or suggestions, please get in touch with us via info@trublo.eu.
Photo by Jackson Simmer on Unsplash
Five Minute Blockchain – No. 50
24.02.2023
Estimated reading time: 6 min 23 seconds
QUOTE OF THE WEEK:
“Everybody says, ‘We don’t want to talk with the banks, we don’t want to know what they’re doing, etc.’ But they’ve actually been around for 300 or 400 years. They have a lot of experience on how to do things actually, or how not to do things.”
Cointelegraph reporting from European Blockchain Convention in Barcelona
TRUST
Top search results could lead to online scammers; FBI recommends ad blockers when using search
Cory Doctorow, a well-known journalist, author and activist, recently published a screenshot of a search via Twitter where the top search result for a restaurant led to a fake website designed to scam users. The actual restaurant was also showing up in the search but ranked lower.
This new approach by scams is hard to detect by Google. As one result, the FBI recommends using an ad blocker to shelter against such falsified and malicious links.
“…cyber criminals are using search engine advertisement services to impersonate brands and direct users to malicious sites that host ransomware and steal login credentials and other financial information.”
Cory Doctorow on Twitter
Malware Bytes Lab
AI-generated voice used to break into a bank account
Many banks offer the option to enter an account using a voice command. A reporter for VICE managed to trick a bank system using a free choice to cheat the system. Experts suggest that banks re-consider and switch to different methods of identity verification.
VICE
Signal warns it might stop services for the UK should “Online Safety Bil” undermine message encrypt.ion
The Online Safety Bill is currently passing through the UK parliament. Boris Johnson, the former prime minister, introduced it. The government and child protection have argued that encrypted messages make it difficult to fight child abuse.
The UK Home Office said in a statement: “It is important that technology companies make every effort to ensure that their platforms do not become a breeding ground for paedophiles. The Online Safety Bill does not represent a ban on end-to-end encryption but makes clear that technological changes should not be implemented in a way that diminishes public safety – especially the safety of children online”.
Meredith Whittaker, president of Signal, told the BBC it was “magical thinking” if government agencies in the UK want to provide privacy, but “only for the good guys. Encryption is either protecting everyone or it is broken for every. one.”
Apple had previously suggested a system where content where photos on phones or tablets could scan for child abuse but had abandoned the plans after much criticism.
BBC
Mozilla study: Developer privacy claims on Google Play can not be trusted
A study by Mozilla calls the labels used in the Google Play Store “a joke” and “useless”:
“The study looked at the privacy information that app developers are supposed to fill out in the Google’s Play Store and compared those details to the apps’ privacy policies. The privacy labels are supposed to give you information about an app’s data practices so you can make informed choices, but the study found the labels are close to useless. Just six apps of the 40 apps in the study got a passing grade.”
Gizmodo
Wired
Using new AI platforms, the next generation of “deepfakes” could leada to a whole new level of damage
The people in the videos look r, and they speak compellingly. But they are not real, and their words might be a believable lie. This is the scenario experts say could be the next level of “deepfakes”.
The combination of several AI platforms “…can also be used to more quickly and cheaply build an army of people who don’t exist, fake actors capable of fluently delivering messages in multiple languages. That makes them useful, says Gregory, for the “firehose” strategy of disinformation preferred by Russia, along with everything from “deceptive commercial personalization to the ‘lolz’ strategies of shitposting at scale.”
Fast Company
It is tempting to use generative AI for legal documents. What could go wrong?
Without the help of technology, lawyers spent hours over hours in legal research and the gradual writing of contracts and other legal documents. Now, generative AI has arrived. And it is quite tempting to use the technology to generate legal texts. On one side, experts consider legal documents to be a good use case for generative AI.
There is already a dedicated an AI platform for this use case called Harvey. Law experts created the company behind Harvey using ChatGPT and received $5 million in funding from OpenAI in November 2022 (TechCrunch). The platform describes itself as a “copilot for lawyers”.
Wired has a story with statements from several prominent law firms. Most are optimistic about the use cases for generative AI, for example, for standard documents or early-stage research. But at the same time, there are worries about current platforms making things up and tending to “hallucinate” about topics poorly defined in the learning material. The significant risk is that some law providers signed not to be more cautious and deploy AI to mass-produce certain legal documents – only to find out later that it was a big mistake.
https://www.harvey.ai (Waitlist)
TechCrunch
Wired ($)
CONTENT
European Union starts consultation on whether some companies should pay more for using internet traffic
The EU considers demanding a contribution from the largest tech companies. The money would then be deployed for upgrades for phone lines both at home and mobile. Currently, a document aims to collect opinions in a survey as public consultation. The EU internal market commissioner, Thierry Breton, will discuss the plans at Barcelona’s Mobile World Congress (MWC). The critical argument that big traffic generators should pay more has been part of lobbying by larger telecommunication companies for some years. Now the EU seems to agree with this view. Opponents are highly critical of such different pricing. They say that the move might open a box of pandora because telcos will be free to draw the line for heavy traffic and enable them to charge all kinds of companies.
EU Commission: Exploratory Consultation – The future of the electronic communications sector and its infrastructure
Thierry Breton on Twitter
TechCrunch
NFT platform Dapper labs to loff off 20% of its staff
Dapper Labs laid off 134 people, or 22% of its staff, in November 2022. Now the company has announced a second round of layoffs. The reduction comes despite a strong financial position. Dapper Labs had received $600 million from venture capital. The company had an early success with NBA Top Shot, where fans could purchase short professional basketball clips.
The Block
YouTube rolls out multi-language audio tracks
“The multi-language audio feature lets creators add dubbing to new and existing videos, helping them expand their global reach and reach new audiences for their channels, according to YouTube.”
Variety
BLOCKCHAIN
International Monetary Fund: Blockchain can speed up payments and settlements, but crypto still is a “disappointment.”
The IMF sees three areas of application: Tokenization, Encryption and programmability. But the advisers argue that private issuers of (crypto) money can not be trusted to protect investors and users.
Blockworks
Research: State of Blockchain Report 2022
Global venture funding provided $26.8B for blockchain companies in crypto finance, web3 and blockchain infrastructure. While prospects looked positive in early 2022, the entire industry came under macroeconomic pressures, specifically in the 4th quarter of the year. CB Insights has a free (after registration) market overview with 162 pages of charts and data to make sense of it all.
CB Insights
SHORT LINKS
- Why the 15-minute city is fueling a ludicrous conspiracy theory (Fast Company)
- Pakistan’s three-day Wikipedia ban sends a “dangerous” message (Rest of World)
- ConsenSys Acquires Easy-to-Use Blockchain Notification Tool ‘Hal’ to Strengthen Web3 Development (Coindesk)
- Tencent to offer “metaverse-in-a-box” development services in Asian markets (Bitcoin.com)
- Google Cloud becomes a validator for Tezos blockchain (Ledger Insights)
Thank you for reading. If you have questions or suggestions, please contact us via info@trublo.eu.
Photo by Vardan Papikyan on Unsplash
Five Minute Blockchain – Nr. 48
10.02.2023
Estimated reading time:
QUOTE(S) OF THE WEEK
“We are grounded in the fact that Google dominates this [search] space. A new race is starting with a completely new platform technology.”
Microsoft CEO Satya Nadella – Interview with “Wall Street Journal” (free)
“If you’ve ever written a blog post or product review or commented on an article online, there’s a good chance this information was consumed by ChatGPT.”
The Conversation
TRUST
How are Microsoft and Google approaching the inclusion of new AI Chatbots in search?
If you want a good overview of how Microsoft and Google approach the race to integrate AI in search: The linked video below compares the presentations from last week by both companies and has further analysis and excerpts from an interview with the Microsoft CEO.
ColdFusion TV: “Google embarrass themselves (AI war is heating up)
So far, Google, the market leader, seems to need more preparation and has not yet recovered from the surprise advance of Open AI and Microsoft’s swift implementation of ChatGPT.
In a promotional video published by Google, users spotted an inaccurate answer. This and a public event which failed to “dazzle” (Reuters) led to a slide of Google stock of 9 % this week; the company lost an equivalent of $100 billion in book value.
Coldfusion TV
Reuters
Hope for the next-gen internet, with new risks and dangers
How AI is integrated into search and other online services is a big question for the coming years. Precisely, what concepts will be followed?
The outcome could be positive: Better search results and better information. Many knowledge jobs could become more productive because the struggle of searching for information could go away. The internet could become a productivity machine, helping us find answers to complex problems.
But there are considerable risks, too. It starts at the foundation: Which material was used for training the new generation of AI platforms? Which pictures, words, insights, concepts, codes and numbers? Who owned those before? How are these info bits used now in AI?
From “The Conversation”:
“ChatGPT is underpinned by a large language model that requires massive amounts of data to function and improve. The more data the model is trained on, the better it gets at detecting patterns, anticipating what will come next and generating plausible text.
OpenAI, the company behind ChatGPT, fed the tool some 300 billion words systematically scraped from the internet: books, articles, websites and posts – including personal information obtained without consent.
If you’ve ever written a blog post or product review, or commented on an article online, there’s a good chance this information was consumed by ChatGPT.”
The Conversation
CONTENT
AI applied to Deep Fakes makes detection much harder
The New York Times has a story about artificial presenters used in videos, which are hard to distinguish from natural persons. According to The Times, an AI software named Synthesia has already been used for Chinese propaganda. The AI-supported software is meant to be used for sales videos. Expect the quality to improve in the coming years.
New York Times (free article)
Quality journalism can be good business
The New York Times published financial results for the 4th quarter of 2022 and some figures for the past year. In short: While traditional advertising is declining, The Times is one news organisation that no longer depends on this type of income. The number of subscribers has grown to 9.3 million, a more significant number than the print circulation even at the height of print publishing.
The New York Times (Free to read)
In addition to hardware sales, Apple’s service business is gigantic
Its service revenue of Apple has reached $78.1B for 2022. That is double the revenue of Netflix ($31.6B) and above the combined revenues of McDonalds and Nike ($72.3B).
FinBold
Macro raises $9.3m for intelligence on top of digital documents
Could standard documents like PDFs become more usable? A company called Macro just got funding for their ideas to do just that.
“Jacob Beckerman, a former investment logic engineer at Bridgewater Associates, grew frustrated using standard document apps like Acrobat and Microsoft Office to print out and mark up documents. He wondered why there wasn’t a way to read and write on a PC that felt as fluid as paper, which led him to experiment with PDF processing software.
By 2020, those experiments had grown into a fully fledged, custom PDF editor that Beckerman helped to build from scratch. Using AI, the editor — called Macro — pulls out key terms, sections and equations to make documents interactive and hyperlinked.”
Link
BLOCKCHAIN
Bank of England published papers on the digital pound
The Bank of England has published two papers discussing a digital pound’s benefits and potential set-up. The current approach favours a central database and does not use smart contracts or blockchain.
In a separate technical paper, blockchain technologies were described that “might have advantages in guaranteeing consistency and resilience”, combined with “privacy, scalability and security challenges. Centrally governed, distributed database technologies might achieve the ledger requirements without limitations. Therefore, these technologies might be appropriate for the core ledger design.”
Coindesk
The Bank of England: Consultation for Digital Pound
The Bank of England: The digital pound. Technology Working Paper
Blockchain Regulation Round-up
- Blockchain privacy at risk in the EU
- Argentina considers “proof of solvency” as a requirement (Bitcoin.com)
- Central Bank of Brazil tests security and transaction privacy levels of Digital Real (Bitcoin.com)
- Coinbase CEO hearing rumours that staking could be banned in the US (Coinmarketcap)
- What public companies are saying about crypto and blockchain on earnings calls (Blockdata)
SHORT LINKS
Thank you for reading. If you have questions or suggestions, please get in touch with us via info@trublo.eu.
Photo by Ales Nesetril via Unsplash
