Showing posts with label AI alignment problem. Show all posts
Showing posts with label AI alignment problem. Show all posts

Monday, August 21, 2023

Five Top Tech Takeaways: SBF Gets Locked Up, Coinbase's Canadian Foray, AI Startups vie with Giants, LK-99 Update, and a Mind-Reading Breakthrough

AI Startups In the Land of Giants

Judge Revokes Bankman-Fried's Bail; Over Witness Tampering Allegations

Sam Bankman-Fried, the founder of FTX, was taken into custody after a judge agreed with federal prosecutors to revoke his bail due to alleged witness tampering. The decision came after a court hearing in New York, and Bankman-Fried was sent to Brooklyn’s Metropolitan Detention Center. Judge Lewis Kaplan rejected Bankman-Fried's plea for delayed detention pending an appeal. The judge believed there was probable cause that Bankman-Fried attempted to tamper with witnesses. Since his arrest in December, he had been on a $250 million bail and was required to stay at his parents' home in Palo Alto, California. The Justice Department accuses him of a "pattern of witness tampering and evading his bail conditions" through his interactions with the media. The defense argued that Bankman-Fried was exercising his first amendment rights. The prosecution's case was strengthened when Bankman-Fried leaked private diary entries of his ex-girlfriend, Caroline Ellison, to the New York Times. Ellison, a former executive of Bankman-Fried’s crypto hedge fund, Alameda Research, had pleaded guilty to federal charges and is cooperating with the government. The prosecution views Bankman-Fried's actions as an attempt to intimidate witnesses indirectly through the media. (Source: CNBC)

Coinbase Embraces Canadian Regulations, Integrates Interac e-transfer

Coinbase is expanding its Canadian operations, integrating Interac e-transfer to simplify transactions in the region and strengthening its presence with over 200 engineers. While other platforms like Binance are retreating from Canada due to tightening regulations by the Canadian Securities Administrators (CSA), Coinbase has embraced these changes, signalling its commitment to the country. The company has complied with the CSA's new rules and has found a positive working relationship with Canadian regulators. Coinbase's CEO Brian Armstrong sees regulatory clarity as a foundation for further growth in the fintech field and remains optimistic about the future of cryptocurrencies globally, even as the firm faces legal challenges in the U.S. (Source: Globe and Mail)

Mind-Reading Breakthrough: UC Berkeley Researchers Vocalize Thoughts.

Researchers at the University of California, Berkeley have made significant progress in the development of devices that can vocalize human thoughts. This advancement could potentially aid patients who have lost their speech abilities due to strokes or brain injuries, allowing them to communicate in a more natural manner. In a notable experiment, the neuroscientists reconstructed Pink Floyd’s song "Another Brick in the Wall, Part 1" using brain activity recordings from 29 patients who listened to the song during brain surgery. While the reconstructed version was not as refined as the original, it was identifiable. The study, which was published in PLOS Biology, demonstrates the potential of using brain-activity patterns to develop therapeutic technologies. Dr. Edward Chang, a neurosurgeon not involved in the study, highlighted the significance of the findings. The research aims to utilize this technology to create neural prosthetics that can restore natural speech abilities to patients. The algorithms developed were even able to reproduce partial vocals from the song. The choice of the Pink Floyd song was due to its balance of familiarity. The breakthrough raises questions about mental privacy, as the ability to interpret thoughts could be the next frontier in privacy concerns. (Source: WSJ)

The Quest for a Room-Temperature Superconductor Continues

The LK-99, initially believed to be a room-temperature superconductor, appears to have different properties than initially thought. Recent studies suggest that in its pure form, LK-99 behaves more like an insulator. This discovery came after the Quantum Energy Research Centre in Seoul, South Korea, shared their initial findings with great enthusiasm. The team had observed certain characteristics in LK-99 that resembled those of superconductors, such as partial levitation above a magnet and a notable drop in electrical resistance. While the initial findings were shared on a preprint server, which allows for rapid dissemination of research without peer review, it's evident that the team was genuinely excited about their discovery, even if subsequent studies have provided a different perspective. (Source: TechCrunch)

Big Tech's Dominance in AI Policy Discussions: Where Do Startups Stand?

In the rapidly evolving realm of generative AI, major players like Microsoft and OpenAI often dominate the conversation, especially when it comes to regulatory discussions. These industry giants have been at the forefront, engaging with policymakers and even entering agreements with the White House to promote responsible AI. However, there's growing concern that smaller AI entities, both commercial and non-commercial, are being overshadowed in these crucial discussions. While these larger companies are instrumental in shaping potential AI policies, smaller businesses, which also play a significant role in the AI ecosystem, are anxious about their limited influence on the outcomes of these regulations. Experts emphasize the importance of including a diverse range of stakeholders in the regulatory dialogue to ensure a balanced and inclusive approach to AI governance. (Source: TheVerge)

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist that is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else. This post was written with the assistance of an AI language model. The model provided suggestions and completions to help me write, but the final content and opinions are my own.

Wednesday, July 26, 2023

Five Top Tech Takeaways: Twitter $20 Billion Brand Bonfire, No Bard for Canada, Apple's GPT and AI Regulations

 

Bonfire of Billions

Musk's Twitter Rebrand: Lighting Up $20 Billion in Brand Value?

Elon Musk's recent decision to rebrand Twitter as "X" and eliminate the iconic bird logo has sparked controversy and is estimated to have wiped out between $4 billion and $20 billion in brand value. The move, which includes a shift in focus towards audio, video, messaging, payments, and banking, has been criticized by analysts and brand agencies who argue that Twitter's brand recognition and cultural influence are invaluable assets. The rebranding has also led to a significant drop in advertising revenue, with advertisers wary of Musk's controversial persona. Despite the backlash, some believe that Musk's personal brand may be powerful enough to carry the new "X" platform forward. (Source: BNN)

Google's Bard Expansion: Canada Left Out in the Cold

Google's AI-powered chatbot, Bard, has expanded globally but has notably excluded Canada, along with countries like China, Russia, Iran, North Korea, Afghanistan, Belarus, and Cuba. This move comes amidst Google's ongoing dispute with the Canadian government over the Online News Act, which mandates tech giants like Google and Meta to negotiate compensation deals with media outlets. The Act aims to balance online advertising revenues, a sector dominated by Google and Meta. In response to the Act, both companies have threatened to block news links from their platforms in Canada. Google's Bard, now available in over 40 languages and more than 230 countries, has not clarified if its exclusion of Canada is directly related to these regulatory disputes. (Source: CTV)

Sam Altman's Eyeball Scans: A New Frontier in Crypto or Privacy Breach?

Worldcoin, a project by OpenAI CEO Sam Altman, has launched a global initiative offering free cryptocurrency in exchange for an eyeball scan to create a digital ID. The project aims to establish a new "identity and financial network" and to verify users as human, not bots. Despite privacy concerns, people in countries like Britain, Japan, and India have participated, with Worldcoin claiming to have issued IDs to over two million people in 120 countries. Critics have raised concerns about potential privacy breaches, but Worldcoin insists that the project is "completely private" and that biometric data is either deleted or stored encrypted. The promise of free cryptocurrency has attracted many participants, despite the potential risks. (Source: CTV)

Apple's AI Ambitions: The Birth of 'Apple GPT'

Apple is reportedly developing its own AI-powered chatbot, internally referred to as "Apple GPT", using a large language model (LLM) framework named "Ajax". The project, which runs on Google Cloud and is built with Google JAX, is still in its early stages with no confirmed plans for public release. Multiple teams within Apple are working on the project, including addressing potential privacy issues. Despite Apple's relative silence in the generative AI space, the company has been integrating AI into its software for years, most notably with Siri. Apple's AI initiative is led by John Giannandrea and Craig Federighi, and a significant AI-related announcement is expected from the company next year. (Source: TheVerge)

AI Giants Commit to New Safety Measures Amid White House Initiative

In an effort to manage the risks associated with artificial intelligence (AI), the Biden administration has reached an agreement with seven major AI companies, including Amazon, Google, Meta Platforms, Microsoft, and OpenAI. The companies have voluntarily committed to implementing more safeguards around AI, such as developing a watermarking system to help users identify AI-generated content, testing their AI systems' security and capabilities before public release, investing in research on the technology's societal risks, and facilitating external audits of system vulnerabilities. While these commitments largely reflect existing safety practices, they lack enforcement mechanisms. The White House is also developing an executive order to govern the use of AI, emphasizing that these commitments are not a substitute for federal action or legislation. (Source: WSJ)

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist that is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else. This post was written with the assistance of an AI language model. The model provided suggestions and completions to help me write, but the final content and opinions are my own.