- Work With AI
- Posts
- Google is Having Employees Rewrite Bard Responses
Google is Having Employees Rewrite Bard Responses
02-17-2023
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/5bf81582-6f83-428a-b2e7-ee351e242df3/02-17-2023.jpg?t=1676608908)
⚡ Today’s Highlights
📰 News: Bing needs some fine tuning + Google employees rewrite AI responses
💰 Funding: NLP automation and AI catering
🦾 Tools: AI models for business data and an AI assistant for reading research papers
🧠 Resources: The Ultimate ChatGPT Guide
📅 Events: Data Science Salon Austin is next week!
📰 Today's Top Stories
(10 min read) (Source: CNBC)
TLDR: Google has asked its employees to help refine its artificial intelligence search tool, Bard, by rewriting answers on topics they understand well. The goal is to improve the accuracy of the tool's responses. After the financially disastrous mistake it made in its launch, Google is leaning on the knowledge of its human employees to clean up the AI's mistakes.
Prabhakar Raghavan, Google's vice president for search, has provided a do's and don'ts page with instructions on how employees should fix responses as they test Bard internally. Google has also set guidelines for rewriting responses to keep them polite, casual, and approachable. Employees have been instructed to maintain an unopinionated, neutral tone and avoid making presumptions. This town just ain’t big enough for two sassy chatbots.
(7 min read) (Source: Bing Blog)
TLDR: Microsoft launched AI-powered Bing just over a week ago. The product is currently in limited preview and is being tested with a select group of people in over 169 countries. Positive feedback from the tests has shown increased engagement across traditional search results and with new features, such as summarized answers and content creation tools.
Most feedback on the AI-generated answers provided by Bing has been positive, with 71% of users giving them a "thumbs up." Microsoft has received good feedback on how to improve the product, and the only way to make improvements is through real-world usage. Bing learns from the real world while maintaining safety and trust, and the search engine has been praised for its ability to provide citations and references to support its answers. The chat feature has also been successful, with users finding it easy to use and approachable.
However, long chat sessions can confuse the AI model and lead to repetitive or unhelpful responses. Microsoft may add a tool to easily refresh the context or start from scratch to address this issue. The AI model can also provide responses that sometimes reflect the tone of the user, unintentionally leading to responses with an inappropriate tone.
The Big Picture: We’ve given Bing our fair share of flack for the crazy things that it’s been saying. But to be completely fair, the potential benefits seem to be outweighing the erratic nature it sometimes adopts. As stated in the blog, the more peculiar responses are a result of longer chat sessions but this will likely get better as it becomes open to the public and receives more data. It’s surprising however that Bing didn’t inherit some of the more diplomatic tendencies of it’s OpenAI cousin, ChatGPT. Only time can tell whether Bing becomes a little less crazy.
(5 min read) (Source: Wall Street Journal)
TLDR: GE Healthcare is investing in developing AI-enabled products to support doctors in analyzing patient data to make more informed treatment decisions. The company is primarily focusing on improving medical imaging by developing algorithms that can detect and diagnose conditions more accurately and efficiently.
Additionally, GE Healthcare is applying AI to hospital operations by creating software that helps hospitals manage patient flow, predict demand and utilization of beds, and improve operational efficiency. The company is also exploring ways to enhance the overall patient experience by developing AI-powered chatbots to schedule appointments, provide test results, and answer healthcare-related questions, among other uses. These efforts aim to improve healthcare outcomes, increase efficiency and accuracy in diagnosis, and enhance patient satisfaction.
(6 min read) (Source: ZoomInfo)
TLDR: ZoomInfo has announced the integration of GPT technology into its platform to help sales and marketing teams find and connect with their ideal customers. The integration will allow users to surface relevant insights at the right time by telling the platform what they are looking for. The combination of generative AI and ZoomInfo’s trusted data will provide real-time insights on who to engage with, when to engage with them, and what to say.
ZoomInfo plans to use GPT technology to shorten cold emails, isolate a call to action from a sales call, and power advanced features like suggested contacts and automation across engagement channels. GPT has already been incorporated into ZoomInfo’s go-to-market plays, and the company expects that the integration will transform the way sales and marketing teams prospect and find customers.
(7 min read) (Source: VentureBeat)
TLDR: Many businesses struggle with demand forecasting, resulting in inventory errors and supply chain management issues. Legacy forecasting tools rely on old and inefficient methods, limiting sales forecasts and leading to overcorrections. However, AI-driven demand sensing has been proven to reduce inventory errors by up to 50%.
AI can draw from multiple datasets to sense demand in real-time, resulting in more accurate forecasting. Additionally, the quality of data is more critical than quantity, and businesses need to take immediate action to improve demand forecasting.
Some businesses have been deterred from adopting AI-driven demand sensing because of the myth that it requires an expensive digital transformation with consultants and data scientists. However, the reality is that businesses already have the data needed to address these issues.
AI tools for demand forecasting will help businesses reduce inventory errors, leading to increased sales and profits. By utilizing AI to improve demand planning, businesses can benefit from more accurate forecasting and better supply chain management.
💰 Funding Alerts
Apkudo, a company specializing in supply chain automation for connected devices, has raised $37.5 million in a Series C funding round. The funding was co-led by Closed Loop Partners' Leadership Fund and Piper Sandler Merchant Banking, with participation from new and existing investors. Apkudo plans to use the funds to expand its commercial and technical operations and international presence, while also reducing e-waste and maximizing the value of connected devices using its AI-powered Circular Industry Platform.
Kern AI, a German startup, offers a platform for NLP developers and data scientists to automate and orchestrate the labeling process for training data. Their ecosystem consists of four products, including standardized code snippets for building NLP automation. Seedcamp, led by managing partner Carlos Espinal, and Faber co-led the seed funding round of €2.7 million. The company plans to expand its platform capabilities and use case catalogue, grow its developer community, and make the platform available to the public.
Axion Ray, a San Francisco-based company, has raised $7.5 million in seed funding from Amplo, Inspired Capital, Boeing, and Tinicum Venture Partners. The company's AI-powered proactive intelligence platform extracts actionable insights from unstructured manufacturing data in real-time, and it plans to use the funds to hire more engineering and AI talent.
Zerocater, a company that The company uses AI and machine learning technology to offer personalized menus based on employee preferences, has raised $15 million in Series C funding led by Cleveland Avenue LLC and Remus Capital. Zerocater intends to use the funds to expand its solutions to 10 new markets, including its first international market in Canada.
🦾 Trending Tools
Patterns is an AI-driven platform that streamlines the process of connecting AI models like ChatGPT to business data and enables users to create a Slack bot to answer customer queries using their existing documentation, and to add AI assist features to their products. With Patterns, users can swiftly progress from idea to full-fledged automation within minutes. The platform also offers a diverse marketplace of data sources and actions, in addition to custom Python and SQL execution, further enhancing its capabilities.
Check out researchGPT if you’re tired of slogging through journal articles. It’s an open-source research assistant that allows you to have a conversation with a research paper or any pdf. The live demo had to be taken down because the creator couldn’t afford the API costs anymore due to its popularity but you can run it locally for yourself by using the code on GitHub!
The 10Web AI Assistant for WordPress is an all-in-one solution for generating, editing, and optimizing SEO content using AI. With just one click, users can create and rephrase SEO-optimized content within the Gutenberg and Classic Editor, while the AI Assistant SEO Pack helps fix errors directly within Yoast SEO. The product saves time by automating SEO error fixes and delivers higher rankings, along with offering 20+ content templates and a document editor. Write and perfect SEO content with the help of AI.
🌎 Popular Content
1. Some people REALLY don’t like how sassy Bing has been since it launched. So much so that there’s a petition floating around to have it shut down. (Twitter)
WTF?? People are literally going crazy... What's next request - arrest AI researchers??
twitter.com/i/web/status/1…— Irina Rish (@irinarish)
6:28 AM • Feb 16, 2023
2. ChatGPT, on the other hand, has been raising concerns about its political biases. After conducting a political compass test on ChatGPT, a researcher created a modified version (and for a surprisingly affordable price) that was fine-tuned to exhibit opposing political values. Thus, RightWingGPT was born. (Twitter)
1. Behold RightWingGPT. An AI model fine-tuned to manifest the opposite political biases of ChatGPT (i.e. to be right wing). Let me describe how I did it and the dangers of politically aligned AIs given their potential to induce societal polarization davidrozado.substack.com/p/rightwinggpt
🧵— David Rozado (@DavidRozado)
9:27 AM • Feb 16, 2023
3. “Johannes Gutenberg trying to save his printing press from a mob of angry scribes”. AI Gutenberg kind of look like a mix of Marv from Home Alone and Jim Carrey. (Reddit)
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/9a10951d-503f-49f0-8ba4-fd179ce4ccb6/image.png)
4. Midjourney is an incredible tool but it makes some horrifying stuff sometimes. Darth Homer might be part Wookie. (Reddit)
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/d30ef004-9592-4b4f-bbfc-c46d872e4f50/image.png)
5. TED-Ed shared this interview with AI expert Stuart Russell in this wonderfully animated video. Stuart shares his opinions on how AI will change the world. (YouTube)
👀 More Reading
Generative AI Can Help You Tailor Messaging to Specific Audiences (Harvard Business Review)
What’s inside a deep neural network (Popular Science)
AI Cameras Make Huge Leap in Protecting the Rarest Animals on Earth (PetaPixel)
🧠 Resources
The Ultimate ChatGPT Guide: More than 100 resources to help you learn how to use ChatGPT to enhance your life. Self-explanatory. Super useful.
Dataversity offers a short plain language explanation to help understand the difference between machine learning and predictive analytics.
📅 Upcoming Events
Data Science Salon Austin (February 21-22, 2023 / Austin, TX + Virtual) is a two-day 500-person conference focused on AI and machine learning applications in the enterprise. The intimate event curates data science sessions to bring industry leaders and specialists face-to-face to educate each other on innovative new solutions in artificial intelligence, machine learning, predictive analytics and acceptance around best practices.
The Data Analytics Career Summit (February 24, 2023 / Virtual) seeks to help individuals break into and move up in the data & analytics fields. The focus of this event will be on helping individuals navigate the many career paths available, such as data analytics, data science and data engineering. Data will also be shared on the current jobs market and the latest survey of individuals in the field. Additional content will focus on best practices for career management, navigation of education options and advice/inspirational stories from real professionals in the field. Registration is now open.
💼 Jobs
Strava: Machine Learning Engineering Intern
Remote, USA
Internship
Roblox: Senior Machine Learning Engineer - Economy
San Mateo, CA, USA
$267,720 - $331,640 USD
Zoox: Senior/Staff AI Software Systems Engineer
Foster City, CA, USA
$180,000 - $280,000 USD
Flagship Pioneering, Inc.: Scientist/Senior Sci Computational Biology -ML/AI
Cambridge, MA, USA
Faire: Staff Machine Learning Engineer & Technical Lead - Search & Recommendation
USA
California / New York
$221,000 - $304,000 USD
Colorado / Washington / New Jersey
$199,000 - $274,000 USD
That's all for todays email! If you want more please follow us at the social channels linked below, or check out our website!
How'd you like today's email? |
Share our newsletter: If you like our work please share/forward this email with your friends, colleagues, and family. It's the best way to support us!
If this email was forwarded to you please sign up here to continue receiving them.
Want your content, product, jobs, or event featured in our newsletter? Reply to this email with the details, and our team will reach out to you.
Do you use AI for work? Tell us how, and you could be featured in our newsletter!
Check out our website for more resources, including a list of AI investors, products, events, and twitter follows.
For an archive of all our posts, click here.
We'd love to hear from you! You can always leave us comments or feedback by replying to this email!