AUGUST 29 HIGHLIGHTS

CEO struggles to keep company afloat after firing 900 employees on a Zoom call in favor of AI

We also look at:

• ​52% of Americans say they’re concerned about the growing use of AI: Pew​, while ​another finding shows that relatively few Americans actually use (or fear) ChatGPT​
• ​AI Pedestrian Detection Systems Have Age and Race Biases, Study Finds​
• ​The US Air Force wants $5.8 billion to build 1,000 AI-driven unmanned combat aircraft​
• Plus: more AI tools, news, and job/career upskilling advice

TOP HEADLINES

Society

​AI could 'turn good jobs into bad jobs'—3 labor historians on what the future of work might hold​

The article explores the potential impact of generative AI tools like ChatGPT on the future of work. Labor historians and experts weigh in on how AI could transform jobs, particularly in higher-paying fields. One concern is that AI could break down complex roles into simpler, lower-paying jobs. For example, a writer's role could be divided into multiple specialized tasks, each requiring less skill and offering lower pay. Another scenario is the complete eradication of certain jobs, as seen historically with the introduction of talkies in the film industry. Despite the challenges, new job roles are emerging, such as AI consultants and AI video editors.

​52% of Americans say they’re concerned about growing use of AI: Pew​

A recent Pew Research Center poll reveals that 52% of Americans are more concerned than excited about the growing use of artificial intelligence (AI). This marks an increase from last year when 38% expressed similar concerns. The survey highlights that while AI has potential benefits in various sectors like healthcare and transportation, there is increasing scrutiny from public officials and a call for regulation. Companies like Nvidia and Microsoft, which are heavily invested in AI, have seen their shares soar, but the public's apprehension remains.

​Early Intervention: Leveraging Visual AI to Prevent Active Shootings​

The article discusses the use of Visual AI to proactively detect threats in schools and prevent active shootings. Traditional methods like gunshot detectors and alarms are reactive and have limitations. Visual AI, developed by SparkCognition, uses computer vision and deep learning to detect suspicious behaviors, such as loitering or unidentified vehicles. The technology integrates with existing camera systems and can send real-time alerts to authorities. The article emphasizes that Visual AI could be a game-changer in ensuring school safety.

​Survey finds relatively few Americans actually use (or fear) ChatGPT​

A recent Pew Research survey reveals that only about 18% of Americans have ever used ChatGPT. The usage varies by demographic, with men, those aged 18-29, and college-educated individuals being more likely to have used the system. The survey also found that people primarily use ChatGPT for entertainment or learning rather than for work. Only 19% of employed respondents who had heard of ChatGPT thought it would significantly impact their jobs. Interestingly, concern about the role of AI in everyday life has risen to 47%, up from 31% last year.

​AI Pedestrian Detection Systems Have Age and Race Biases, Study Finds​

A study by King’s College London reveals significant age and race biases in pedestrian detection systems used in autonomous vehicles. The systems were 20% more accurate in detecting adults than children and 7.5% more accurate in detecting light-skinned pedestrians over darker-skinned ones. These biases are attributed to unrepresentative training data and could pose safety risks, especially during nighttime driving. The study calls for more transparency and tighter regulations in pedestrian detection systems.

Technology

​Advances in AI, Chips Boost Voice Recognition​

IBM and California universities at San Francisco and Berkeley have made significant advancements in speech recognition technology. IBM has developed a more energy-efficient chip that turbo-charges speech-recognition models. The chip incorporates phase-change memory devices, optimizing AI processes and reducing energy consumption. Meanwhile, researchers at UC San Francisco and UC Berkeley have created a brain-computer interface that can generate words from a user's thoughts, offering promise for patients suffering from vocal paralysis and speech loss.

​Poe’s new desktop app lets you use all the AI chatbots in one place​

Poe, an AI chatbot platform by Quora, has rolled out several updates including a Mac app and an enterprise tier. The platform aims to be a one-stop-shop for AI chatbots, allowing users to interact with multiple bots like ChatGPT-4 and Anthropic's Claude. For a subscription fee of $19.99 per month or $200 per year, users get unlimited access to all bots on the platform. The company is also planning to introduce an enterprise tier for businesses.

​Sexy AI Chatbots Are Creating Thorny Issues for Fandom​

The article delves into the complexities surrounding AI chatbots like Character.AI in the realm of fandom. Character.AI allows users to interact with fictional characters, but the platform's rise has sparked debates about copyright, adult content, and the nature of fan interaction. While some see these chatbots as a new form of fan engagement, others worry that they could isolate fans and raise legal issues. The platform has over 15 million registered users and has been met with both enthusiasm and skepticism.

​AI could choke on its own exhaust as it fills the web​

The article discusses the risks associated with the increasing amount of AI-generated content on the internet. Experts predict that up to 90% of online content could be AI-generated in the near future. This poses challenges for both human society and AI systems. For humans, the problem lies in information overload and the potential loss of jobs in content creation sectors. For AI, the issue is "model collapse," where AI models trained on data produced by other AIs can deteriorate in quality. The article suggests that companies should preserve access to pre-2023 data to avoid being affected by model collapse.

​Fruit Growers' AI Tool Launches in Cambridge After £28M Fundraise​

FruitCast, a digital forecasting tool for soft fruit growers, is set to launch in 2024 after securing £2.8 million in funding from private investors, Ceres Agritech, and Innovate UK. Developed as a spin-out from Ceres Agritech, the tool aims to help growers more accurately forecast fruit growth, thereby aiding in labor planning and market supply. The technology uses a moving camera to take thousands of photographs, measuring individual fruit numbers, weight, and maturity.

CEO Richard Williamson and CTO Dr. Raymond Martin lead the project. The tool has been tested successfully with two of the UK's leading soft fruit growers, W B Chambers and Place UK. FruitCast will offer its services through a subscription model and aims to expand its capabilities to other crops in the future.

The technology is designed to address challenges in the industry, such as labor costs, weather variations, and competitive imports. It aims to improve efficiency without replacing human labor, complementing the skills of on-site workers. Initial funding of £520,000 in 2022 allowed for the tool's development, and a recent investment of over £1.1 million will take it to launch.

​Scientists Design an AI To Make Computing Thousands of Times Faster​

Computer scientists at the University of Massachusetts Amherst have developed a tool called Scalene to improve the efficiency of Python, a popular but notoriously slow coding language. Led by Professor Emery Berger, the team created Scalene to identify and fix inefficient code segments. The tool uses generative AI to suggest optimized code after identifying areas where the program is wasting time. Berger emphasizes that future speed improvements in computing will come more from efficient programming than from hardware advancements. Scalene is available for download on GitHub.

​A Study on Robustness and Reliability of Large Language Model Code Generation​

The study by Li Zhong and Zilong Wang examines the reliability and robustness of code generated by Large Language Models (LLMs) like GPT-4. It highlights that while LLMs are increasingly consulted by software engineers for coding questions, the generated code often contains API misuses. These errors could lead to severe issues like resource leaks and program crashes. The study introduces a dataset, RobustAPI, for evaluating the reliability and robustness of code generated by LLMs. The evaluation reveals that even for GPT-4, 62% of the generated code contains API misuses, posing risks if applied in real-world software.

Education

​Not just "messing with a robot": Georgia school district brings AI into classrooms, starting in kindergarten​

Gwinnett County School District in Georgia has integrated AI into its curriculum, starting from kindergarten. The initiative aims to prepare students for a future where AI will be ubiquitous. The curriculum is not limited to computer science but extends to subjects like English and art. Students are engaged in creative problem-solving, coding, and even using AI tools for data analysis in social studies. The district's K-12 program, called "Computer Science for All," has received positive feedback from students, parents, and teachers.

​'It's not that accurate': Teachers rely on apps to catch students using artificial intelligence​

Teachers are grappling with the rise of students using AI to cheat on assignments. Apps like "Turnitin" are employed to detect AI-generated content, but their effectiveness is inconsistent. Dr. Guy McHendry at Creighton University conducted tests and found that these apps sometimes fail to catch AI-generated work. The article also highlights that students are using multiple apps to humanize the tone of AI-generated content, making detection even more challenging. Some educators are reverting to traditional methods like handwritten assignments to counteract cheating.

Security and Policy

​DOD Will Deploy AI-Enabled Detection System to Monitor D.C. Airspace​

The U.S. Department of Defense (DoD) is set to deploy an AI-powered airspace monitoring system to enhance the protection of Washington, D.C. The system, developed in partnership with the Defense Innovation Unit, will replace the existing 9/11-era system and offer a tenfold increase in performance. Teleidoscope, a first-time DoD vendor, has been awarded a $100 million contract for the system, which will begin fielding this year. The technology aims to improve air defense capabilities and has broad applications for national defense against threats like unmanned aerial systems and cruise missiles.

​How PayPal is using AI to combat fraud, and make it easier to pay​

PayPal is leveraging artificial intelligence to enhance its security measures and user experience. John Kim, the company's Chief Product Officer, discussed how AI is helping to detect complex fraud patterns and anticipate changes in fraudulent activities. PayPal is planning to launch three new AI-based products in the next 120 days, including a checkout feature that predicts the correct address to use with a specific merchant. While AI offers promising applications, Kim also expressed caution about over-relying on the technology, especially when human customer service agents could perform certain tasks more effectively.

​The US Air Force wants $5.8 billion to build 1,000 AI-driven unmanned combat aircraft​

The U.S. Air Force has requested $5.8 billion to develop AI-driven collaborative combat aircraft. These autonomous aircraft are designed for high-risk missions and to protect human pilots. The XQ-58A Valkyrie aircraft is a potential candidate for this program and will be tested later this year in a simulation over the Gulf of Mexico. While the military sees this as a step forward, human rights advocates warn that outsourcing killing to machines crosses a moral boundary and could lead to rapid conflict escalation.

​AI is biased. The White House is working with hackers to try to fix that​

The White House collaborated with hackers at the Def Con convention to identify and address biases in AI. The event was the largest-ever public red-teaming challenge, where hackers probed AI technology for biases. The White House urged top tech companies like Google and OpenAI to have their models tested by independent hackers. The challenge revealed that while some AI models could answer questions appropriately, they still exhibited biases when prompted with specific scenarios. The findings will be reviewed over the next few months, allowing tech companies to engineer their products differently to eliminate biases.

​Generative AI Can Lead to Helpful Lemon Law Case Outcomes​

As vehicle sales grow, original equipment manufacturers (OEMs) are facing an increase in Lemon Law lawsuits, particularly in California. The article suggests that OEMs can mitigate the number of Lemon Law cases by adopting generative AI technologies. These AI systems can help in managing litigation more efficiently and matching the capabilities of plaintiff-side law firms.

​Deepfake Video Phone Calls Could Be a Dangerous AI-Powered Scam​

In 2019, a CEO of a British energy company was scammed out of €220,000 ($249,000) when he received a call from a voice that sounded like the head of the company's German parent unit. The voice, generated using artificial intelligence, instructed him to wire money to a Hungarian supplier. Rüdiger Kirsch, a fraud expert at Euler Hermes, the energy company's insurer, confirmed that the voice was AI-generated. The insurer paid out the claim but did not disclose the identity of the client. This incident highlights the evolving tactics of cybercriminals, who are now using AI-generated voice simulation as a new method for fraud. Expect these types of incidents to proliferate as AI gets better.

Business

​OpenAI releasing version of ChatGPT for large businesses​

OpenAI has launched ChatGPT Enterprise, a version of its popular ChatGPT aimed at large businesses. This new offering provides enhanced security, privacy, and faster access to OpenAI's technology. Early adopters include companies like Block, Carlyle, and Estee Lauder. The launch aims to make employers more comfortable with the use of ChatGPT in the workplace, as many employees have been using the consumer version for work-related tasks despite employer discouragement.

Microsoft, a financial backer of OpenAI, also offers business access to ChatGPT through its Azure OpenAI Service. However, ChatGPT Enterprise does not require a subscription to Azure. Both companies have previously introduced overlapping services, and it remains unclear to what extent they are competing for the same customer base. OpenAI states that customers can choose the platform that best suits their business needs.

​Yahoo Mail isn’t dead and adds AI capabilities​

Yahoo Mail is enhancing its platform with new AI features, developed using Google Cloud’s AI platform. One notable addition is the "Shopping Saver" feature, which scans users' inboxes for unused gift cards, discount codes, and store credits, and helps draft messages to utilize these savings. The feature aims to assist users in saving both time and money. Other AI functionalities include an improved search mode, a writing assistant for tone guidance, and a message summary feature that highlights important information in emails.

​Better.com CEO Who Fired 900 People On Zoom Is Struggling To Keep His Company Afloat​

Better.com, once valued at $7.7 billion, is facing a financial crisis. The company went public via a SPAC merger despite having no plans for profitability and a plummeting stock price. CEO Vishal Garg, known for his controversial leadership style, has been criticized for the company's decline. The article highlights multiple issues, including an SEC inquiry, the dismissal of 7,000 employees, and a lack of financial controls. Investors are increasingly skeptical about the company's future, and some have even withdrawn their support.

​Google's Gemini AI is a Serious Threat to ChatGPT – Here's Why​

Google is developing Gemini, a large language model (LLM) that could pose a significant threat to ChatGPT. The model combines techniques from AlphaGo, such as reinforcement learning and tree search, with GPT-4 capabilities. Google's proprietary data from services like Google Search, YouTube, and Google Books could give Gemini a distinct edge in sophistication. The model is expected to be released in the fall of 2023 and will be multimodal, capable of generating both text and images. The article suggests that OpenAI will face stiff competition from Google, especially given Google's extensive AI research capabilities.

​Qualiti.ai Raises $6.5M in Seed Funding​

Qualiti.ai, a Utah-based software test automation solution provider, has secured $6.5 million in seed funding led by Sierra Ventures, with participation from Epic Ventures. The company plans to use the funds to expand its sales and engineering teams. Founded by industry veterans Peter West and Jeff Handy, Qualiti aims to automate software testing without human input. The AI-driven platform can write high-quality automated tests, manage ongoing test authoring, and report bugs in real-time, potentially replacing up to 30% of a company's engineering budget.

Productivity

​Canadian workforce’s surging interest in AI skills revealed in new surveys​

LinkedIn's latest Workforce Confidence Index reveals a growing interest in AI skills among Canadian workers. The report shows that 50% of Canadian workers are keen on acquiring AI-related skills to advance their careers. Over a quarter of professionals reported a 25% growth in AI's role in their work over the past year. While younger generations like Millennials and Gen Z show stronger interest in AI, the report also emphasizes the importance of soft skills. Interestingly, only 5% of Canadian executives intend to reduce headcount due to AI, highlighting a commitment to growth and innovation.

Media and Entertainment

​Major websites like Amazon and the New York Times are increasingly blocking OpenAI's web crawler GPTBot​

OpenAI's web crawler, GPTBot, is facing increasing resistance from major websites like Amazon and The New York Times. The crawler is used to collect data for training ChatGPT. As of this week, 70 of the world's top 1,000 websites have moved to block GPTBot. The blocking is implemented through a file called robots.txt, which GPTBot is programmed to respect. The increasing trend of blocking is attributed to concerns over copyright rules and data ownership. Several lawsuits are already in the works, and there is a growing awareness about the ethical implications of data scraping by AI projects.

​Content Farms Are Using AI Chatbots to Plagiarize News Outlets​

Online content farms are using AI chatbots to rewrite and republish news stories from major outlets like The New York Times and CNN. The report by NewsGuard found 37 such sites that appeared to be fully automated. These plagiarized articles are being used to generate ad revenue from major brands. The report warns that there could be hundreds or even thousands of such sites that have not been identified yet. Both OpenAI and Google have policies against using their models for plagiarism, but enforcement appears to be lax.

​Watch out, Midjourney! Ideogram launches AI image generator with impressive typography​

Ideogram, a new generative AI image startup founded by former Google Brain researchers, has launched with $16.5 million in seed funding led by a16z and Index Ventures. The startup differentiates itself by offering reliable text generation within images, a feature lacking in most other AI image generators. The company's web app provides various preset styles, including typography, 3D rendering, and more. Ideogram aims to appeal to graphic designers and those who need eye-catching text in their images.

Healthcare

​ChatGPT Shows ‘Impressive’ Accuracy in Clinical Decision Making​

A study led by Mass General Brigham found that ChatGPT was about 72% accurate in overall clinical decision-making across all medical specialties and phases of care. The study tested ChatGPT's ability to work through an entire clinical encounter, from initial diagnosis to care management decisions. The AI chatbot performed equally well in both primary care and emergency settings. The study suggests that large language models like ChatGPT have the potential to augment medical practice by supporting clinical decision-making with impressive accuracy.
​

SUPPORT INNOVATION DISPATCH

AI TOOLS TO WATCH

Companies are creating new AI tools nearly every day. There's not enough time for me to use all of them and provide an in-depth analysis, so please don't take their listings here as an endorsement. This is just to show you what's out there and what you can (supposedly) do with them.

​SmartBot‑AI Chatbot - SmartBot - AI-Powered Customer Support Solution for Shopify​

SmartBot is an AI-driven customer support solution designed for Shopify stores. Developed by BestChat, it aims to enhance customer satisfaction, automate support, and boost sales. The platform supports multiple languages and offers features like personalized product recommendations, order tracking, and post-sale services. It is a no-code platform, making it easy to integrate with Shopify.

​Salssy - The #1 LinkedIn SalesBot​

Salssy is an AI-powered LinkedIn Sales Agent designed to automate 100% of your LinkedIn outreach. The platform offers a risk-free 30-day money-back guarantee and requires no prior experience. It aims to save users up to 15 hours weekly by automating LinkedIn chats, allowing them to focus on more productive tasks. Salssy provides tailored conversations, security features, and data enrichment, among other capabilities. It also offers detailed analytics to help users make data-driven decisions.

​Chatwith - Next-Gen AI Chatbot for Your Website​

Chatwith offers a next-generation AI chatbot that does more than just chat. It can be trained on your content and back-end data to handle all customer questions, from real-time inventory information to document generation. The platform is designed to boost customer service and reduce support workload. It offers various features like conversational search, Q&A, and learning assistance. The chatbot can be easily added to popular platforms like Shopify, Wordpress, and Webflow with just a single line of code.

​Profilio - Accelerate Your Recruitment With AI-Powered CV Analysis​

Profilio is an AI-powered platform designed to streamline the recruitment process. It offers features like bulk CV analysis, advanced job matching, and real-time results dashboards. The platform can analyze and rank candidates based on their qualifications, experiences, and skills, saving HR teams time and resources. Profilio is suitable for companies of all sizes and offers various pricing plans.

EMAIL ME ABOUT SPONSORSHIP AND ADVERTISING

RETOOLING

In this section, we look at recent job postings and cover the AI-related skills necessary to do these jobs. This section is intended to shed light on the training or re-skilling that you or your employees need to focus on as the AI revolution moves forward.

Today's job analyses:

​Senior Software Engineer, Full Stack, Java, AWS​

• Company: Capital One (Financial services and banking)
• Location: Paris, Texas
• Compensation: Not specified

AI Impact:

Automation in software engineering is already prevalent, with CI/CD pipelines, automated testing, and code review bots. AI could further automate tasks like debugging, code generation, and even some aspects of design. Machine learning models could analyze vast codebases to suggest optimizations or even write code snippets.

Staying Ahead:

Human engineers can focus on problem-solving, architecture, and design, areas where AI still lags. Soft skills like team collaboration, leadership, and understanding business needs are irreplaceable. Continuous learning about new technologies and methodologies will also keep an engineer relevant.

---

​Field Marketing Manager​

• Company: KForce (Professional staffing and solutions)
• Location: Houston, Texas
• Compensation: $179,020.00 (estimated)

AI Impact:

AI can automate data analysis for customer segmentation, lead scoring, and even content generation. Chatbots can handle customer queries, and programmatic advertising can replace manual media buys.

Staying Ahead:

Creativity and strategy are areas where humans excel. Understanding consumer psychology, building relationships, and strategic planning are skills that AI can't replicate. Also, staying updated with the latest marketing trends and technologies will make a human candidate more valuable.

---

​Nurse Practitioner (NP) - Psychiatry in Fort Smith, AR​

• Company: Locum Jobs Online (LJO) (Medical job listings and staffing)
• Location: Fort Smith, Arkansas
• Compensation: $117,190.00 (estimated)

AI Impact:

AI can handle diagnostic assistance, medication management, and even psychological assessments to some extent. Telemedicine bots can provide basic mental health support, and AI algorithms can analyze medical records for patient history.

Staying Ahead:

Empathy, ethical decision-making, and the human touch are irreplaceable in healthcare. Keeping up-to-date with medical research and technologies, as well as focusing on patient-centered care, will keep a human healthcare provider indispensable.