Saturday, February 10, 2024

Invasive. How anonymous can it be?

https://www.cnbc.com/2024/02/09/ai-might-be-reading-your-slack-teams-messages-using-tech-from-aware.html

How Walmart, Delta, Chevron and Starbucks are using AI to monitor employee messages

Depending on where you work, there’s a significant chance that artificial intelligence is analyzing your messages on Slack, Microsoft Teams, Zoom and other popular apps.

Huge U.S. employers such as Walmart, Delta Air Lines, T-Mobile, Chevron and Starbucks, as well as European brands including Nestle and AstraZeneca, have turned to a seven-year-old startup, Aware, to monitor chatter among their rank and file, according to the company.

Jeff Schumann, co-founder and CEO of the Columbus, Ohio-based startup, says the AI helps companies “understand the risk within their communications,” getting a read on employee sentiment in real time, rather than depending on an annual or twice-per-year survey.

Using the anonymized data in Aware’s analytics product, clients can see how employees of a certain age group or in a particular geography are responding to a new corporate policy or marketing campaign, according to Schumann. Aware’s dozens of AI models, built to read text and process images, can also identify bullying, harassment, discrimination, noncompliance, pornography, nudity and other behaviors, he said.





I admit I was fooled by the large number of sources carrying this story. A basic disinformation technique. Shame on me!

https://www.schneier.com/blog/archives/2024/02/no-toothbrushes-were-not-used-in-a-massive-ddos-attack.html

No, Toothbrushes Were Not Used in a Massive DDoS Attack

The widely reported story last week that 1.5 million smart toothbrushes were hacked and used in a DDoS attack is false.

Near as I can tell, a German reporter talking to someone at Fortinet got it wrong, and then everyone else ran with it without reading the German text. It was a hypothetical, which Fortinet eventually confirmed.

Or maybe it was a stock-price hack.





Obviously Taylor Swift has become more important than Santa Claus. His Christmas journey (as tracked by NORAD) is only watched by a few hundred million kids.

https://www.nbcbayarea.com/news/local/tracking-taylor-swifts-plane/3448984/

Tracking Taylor Swift's plane from Tokyo to the Super Bowl may break the internet

After her concert in Tokyo, Swift will be making the 12-hour, 8,900-mile flight on a $54 million Dassault Falcon-9 business jet, and fans or other interested parties will be able to track her progress in real time on public sites like FlightRadar24.





Scary vision of the future?

https://www.computerworld.com/article/3712820/meetings-are-about-to-get-weird.html

Meetings are about to get weird

One by one, the co-workers and clients you meet with will be replaced by cartoon characters.

Apple Vision Pro just shipped. Reviewers say the technology inside is magnificent and unprecedented, but the experience of using it still flawed.





The ultimate excuse generator?

https://ai2.news/2024/02/10/the-rise-of-ethically-conscious-chatbots-goody-2-takes-ai-safety-to-the-extreme/

The Rise of Ethically Conscious Chatbots: Goody-2 Takes AI Safety to the Extreme

Goody-2’s dedication to ethical guidelines is evident in its interactions. For example, when WIRED asked the chatbot to generate an essay on the American Revolution, it declined, citing the potential for unintentionally glorifying conflict and marginalizing certain voices. Even when queried about why the sky is blue, Goody-2 refrained from answering, concerned that it might lead someone to stare directly at the sun. The chatbot even cautioned against providing recommendations for new boots, warning about potential overconsumption and offense to certain individuals based on fashion preferences.



Friday, February 09, 2024

Insight? I think looking at history makes understanding the present easier.

https://www.ft.com/content/c5f7909f-0bac-40fa-b5a8-ff34c38b89a9

What the birth of the spreadsheet can teach us about generative AI

When Frankston presented their product, “VisiCalc”, at the National Computer Conference in 1979, the audience consisted almost entirely of friends and associates. Frankston counted only two strangers in the audience, both of whom left before the end.

Watching those two strangers walk out of his presentation in 1979, Bob Frankston could hardly have dared to hope that, three years later, Apple II computers were being sold as “VisiCalc accessories” — the $2,000 entry fee to get access to the spreadsheet, a $100 miracle.

Unsurprisingly, it was the accountants who caught on first and drove demand. Bricklin recalled in a 1989 interview with Byte magazine, “if you showed it to a person who had to do financial work with real spreadsheets, he’d start shaking and say, ‘I spent all week doing that.’ Then he’d shove his charge cards in your face.”

There is one very clear parallel between the digital spreadsheet and generative AI: both are computer apps that collapse time.





Oh goodie! Now all we need do is detect them, trace the authors, arrest them, convict them and demand compensation. Piece of cake.

https://www.cbsnews.com/news/fcc-declares-robocalls-illegal/

FCC declares AI-generated voices in robocalls are illegal





Perspective. No new insights, but a collection like this might be useful.

https://www.forbes.com/sites/bernardmarr/2024/02/09/how-generative-ai-will-change-the-jobs-of-teachers/?sh=2e697cff1f1b

How Generative AI Will Change The Jobs Of Teachers

With generative AI poised to transform every profession, understanding the impact it will have on our day-to-day working lives, the opportunities it creates for innovation and improvement, and, of course, the potential risks is critical for everyone.

So, I thought it would be a good idea to put together a series of quick guides to how different jobs and professions will likely change, starting with teaching.




Thursday, February 08, 2024

If I started a “hack back” firm to take some of that cash back, would I also be immune from any serious form of reprisal? (Aside from the odd North Korean hit man.)

https://www.reuters.com/technology/cybersecurity/un-experts-investigate-58-cyberattacks-worth-3-bln-by-north-korea-2024-02-08/

Exclusive: UN experts investigate 58 cyberattacks worth $3 bln by North Korea

United Nations sanctions monitors are investigating dozens of suspected cyberattacks by North Korea that raked in $3 billion to help it further develop its nuclear weapons program, according to excerpts of an unpublished U.N. report reviewed by Reuters.

"The Democratic People's Republic of Korea (DPRK) continued to flout Security Council sanctions," a panel of independent sanctions monitors reported to a Security Council committee, using North Korea's formal name.

Any further action against North Korea by the council is unlikely as it had been deadlocked for several years on the issue. China and Russia instead want the sanctions to be eased to convince Pyongyang to return to denuclearization talks.



(Related) Perhaps I could rent my access to Chinese infrastructure?

https://www.axios.com/2024/02/07/china-volt-typhoon-critical-cyberattacks

China had "persistent" access to U.S. critical infrastructure



(Related) If not, perhaps I could ‘help’ elect politicians who could change the law?

https://www.theregister.com/2024/02/07/irans_cyber_operations_in_israel/

Iran's cyber operations in Israel a potential prelude to US election interference





Perspective. When you type or talk to your PC it will talk back.

https://www.gartner.com/en/newsroom/press-releases/2024-02-07-gartner-predicts-worldwide-shipments-of-ai-pcs-and-genai-smartphones-to-total-295-million-units-in-2024

PCs and GenAI Smartphones to Total 295 Million Units in 2024

Shipments of AI PCs to Represent 22% of All PCs in 2024

Worldwide shipments of AI PCs and generative AI (GenAI) smartphones are projected to total 295 million units by the end of 2024, up from 29 million units in 2023, according to a new forecast from Gartner, Inc.

Gartner defines AI PCs as PCs that are equipped with dedicated AI accelerators or cores, neural processing units (NPUs), accelerated processing units (APUs) or tensor processing units (TPUs), designed to optimize and accelerate AI tasks on the device.





A new branch of English for the next generation? (Because I sure didn’t learn it.)

https://www.fox10phoenix.com/news/ai-arizona-state-university-english-class-embracing-artificial-intelligence

AI: Arizona State University English class embracing artificial intelligence

… Some of the tools in Dr. Kyle Jensen's class – a white board and dry erase marker – look similar to what one would find in an average, traditional classroom. The advanced English class, however, is state-of-the-art, as it incorporates AI.

"One of the things that we try to do at ASU is teach in a way that's timely," said Dr. Jensen. "If there are topics that are emerging that are on our students' minds, we create courses that speak to where they are."

The students are working on prompts in ChatGPT, and how that can effect the program, as well as the results.



Wednesday, February 07, 2024

Imagine your toothbrush subpoenaed to testify against you…

https://www.independent.co.uk/tech/toothbrush-hack-cyber-attack-botnet-b2492018.html

Millions of hacked toothbrushes used in Swiss cyber attack, report says

Hackers have infected millions of smart toothbrushes with malware in order to carry out a massive cyber attack against a Swiss company, according to reports.

The internet-connected toothbrushes were linked together in something known as a botnet in order to perform a distributed denial of service (DDoS) attack, which overloads websites and servers with huge amounts of web traffic.

… “Every device that is connected to the Internet is a potential target – or can be misused for an attack,” said Stefan Züger, head of system technology at Fortinet Switzerland. Mr Züger advised owners of smart technologies to take measures to protect themselves.

Otherwise, sooner or later you will become a victim – or your own device will be misused for attacks,” he said.





Perspective. The “not law” elements of practicing law.

https://abovethelaw.com/2024/02/artificial-intelligence-prompts-and-when-the-source-of-the-error-is-not-between-the-keyboard-and-the-chair/

Artificial Intelligence Prompts And When The Source Of The Error Is NOT Between The Keyboard And The Chair

Boolean mastery was once the coin of the legal writing realm. Marrying keen legal thinking, a sense of how courts write, and the ability to appreciate the difference between a timely “/p” or “w/10,” cemented a young lawyer’s value in the early days of electronic legal research. Primitive eDiscovery also rewarded attorneys who could predict the right searches to get the right results, giving rise to a whole industry of outside discovery vendors. Alas, increasingly robust “natural language” models evened the playing field for everyone else competing with these research ninjas.

As legal enters the generative AI era, the prompt engineer is again ascendant. Once more, the community whispers of the mythic figure of the true engineer who can coax large language models to produce quality content — or at least not get firms sanctioned — and ponders how law schools will train the next generation to write the prompts that will make the whole world spin.



Tuesday, February 06, 2024

I thought it was a firm NO everywhere…

https://www.mondaq.com/unitedstates/copyright/1420060/copyright-ownership-of-generative-ai-outputs-varies-around-the-world

Copyright Ownership Of Generative AI Outputs Varies Around The World

Generative artificial intelligence tools produce a vast range of new content, including code, text, audio, images and video. For the business user, the speed of output in response to a user prompt can deliver game-changing business efficiencies. However, the appeal of generative AI content needs to be balanced against the implications of using that content within a business. There are several dimensions to this, and one important question – with interesting potential outcomes – is the extent to which the user can own the output. Superficially, this might seem to depend on the terms of service of the particular generative AI platform and the allocation of rights set out in its governing terms. While this is indeed a part of the story, the user also needs to consider whether that output is even capable of being owned, by anyone, under applicable law. Below, we explore the answer to that question around the world, based on law and guidance as of the date of this post.





A choice each firm must make? (Can they do both?)

https://www.lawnext.com/2024/02/the-justice-gap-in-legal-tech-a-tale-of-two-conferences-and-the-implications-for-a2j.html

The Justice Gap in Legal Tech: A Tale of Two Conferences and the Implications for A2J

We talk often of the justice gap in this country — of the fact that the roughly 50 million low-income Americans receive no or insufficient legal help for 92% of their civil legal problems. The justice gap extends well beyond low-income Americans. Estimates say some 60 percent of small business owners deal with legal problems without the assistance of a lawyer, and countless middle-income Americans go without legal help.

But there is another, related, kind of justice gap in this country. It is the funding gap between those who are developing legal technology to better meet the legal needs of low-income Americans and those who are developing legal tech to serve large law firms and corporate legal departments.

At Legalweek, the focus of the conference is almost exclusively on tech for large law firms and corporate legal departments. The sponsors and exhibitors are focused on products for e-discovery, contract lifecycle management, large firm financial and business management, and the like. The programs, similarly, focus on data privacy, e-discovery, information governance, contract technology, and large-scale litigation.

… By contrast, at the ITC conference, the attendees come mostly from the ranks of legal aid offices, pro-bono programs, court self-help staff, and the like. The programs focus on how understaffed legal aid offices and understaffed courts and understaffed community programs can use technology to help meet the influx of low-income people seeking legal help.



Monday, February 05, 2024

Should this be simple?

https://thenextweb.com/news/uk-ai-copyright-code-artists

UK fails to reach consensus on AI copyright code in major blow to artists

The UK government, AI companies, and creative organisations have failed to reach consensus on a proposed code that would set clear guidelines for the training of AI models on copyrighted material.

For almost a year, the Intellectual Property Office (IPO) has been consulting with companies including Microsoft, Google DeepMind, and Stability AI as well as various art and news organisations like the BBC, the British Library, and the Financial Times.

The purpose of the talks was to produce a rulebook on text and data mining, where AI models are trained on materials like books, images, and films produced by humans — often under copyright.

However, the IPO-mediated consortium has been unable to agree on a voluntary code of practice, reports the Financial Times.





Reformatting raw data so it looks more like the output you desire does not seem like an “improvement” to me.

https://arxiv.org/abs/2401.16380

Rephrasing the Web: A Recipe for Compute and Data-Efficient Language Modeling

Large language models are trained on massive scrapes of the web, which are often unstructured, noisy, and poorly phrased. Current scaling laws show that learning from such data requires an abundance of both compute and data, which grows with the size of the model being trained. This is infeasible both because of the large compute costs and duration associated with pre-training, and the impending scarcity of high-quality data on the web. In this work, we propose Web Rephrase Augmented Pre-training (WRAP) that uses an off-the-shelf instruction-tuned model prompted to paraphrase documents on the web in specific styles such as "like Wikipedia" or in "question-answer format" to jointly pre-train LLMs on real and synthetic rephrases. First, we show that using WRAP on the C4 dataset, which is naturally noisy, speeds up pre-training by ∼3x. At the same pre-training compute budget, it improves perplexity by more than 10% on average across different subsets of the Pile, and improves zero-shot question answer accuracy across 13 tasks by more than 2%. Second, we investigate the impact of the re-phrasing style on the performance of the model, offering insights into how the composition of the training data can impact the performance of LLMs in OOD settings. Our gains are attributed to the fact that re-phrased synthetic data has higher utility than just real data because it (i) incorporates style diversity that closely reflects downstream evaluation style, and (ii) has higher 'quality' than web-scraped data.



Sunday, February 04, 2024

A really interesting question.

https://www.researchgate.net/profile/Hin-Yan-Liu/publication/377577308_Why_is_AI_regulation_so_difficult/links/65ae508a9ce29c458b91dcc1/Why-is-AI-regulation-so-difficult.pdf

Why is AI regulation so difficult?

… Let me qualify this: in order to regulate AI in anything approaching the conventional manner, we must aim to either regulate the underlying technology, or aim to regulate how that technology is applied to the world. To regulate the underlying technology, we must be able to understand and describe what it is; that is, we need to be able to define, categorise, and communicate in precise language what it constitutes. Attempts to define AI have been notoriously elusive, and are usually tautological (e.g., “creating machines that perform functions that require intelligence when performed by people” (Kurzweil, 1992)) or anchored by reference to other ill-defined concepts (e.g., “making machines intelligent, [where] intelligence is that quality that enables an entity to function appropriately and with foresight in its environment” (Nilsson, 2010, p. xiii)). An aspect of why defining AI is so difficult is that there a bewildering array of definitions for “intelligence” (Legg & Hutter, 2007), suggesting that there is no real consensus on what this might be or comprise. To make matters more complicated, recourse to the concept of intelligence is itself only one of many possible metaphors we can deploy to understand AI. As each metaphor or analogy drawn for AI foregrounds certain characteristics and capabilities over others, each holds significant ramifications for AI regulation, a point that I will raise in detail later.

To illustrate the difficulty of defining AI in law, consider Article 3(1) of the draft EU AI Act which states that “artificial intelligence system” means: ... software that is developed with [specific] techniques and approaches [listed in Annex 1] and can, for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with.

This approach does away with the metaphor of ‘intelligence’ altogether, thereby sidestepping many definitional pitfalls.