The most charming trick is the largest Flaw
How generative AI makes people feel like they are using magic: David Song, founder of Stability AI, and his list of startup ideas for generating language and text
Generative AI enthusiasts predict the technology will take root in all kinds of industries and will do much more than just spit out images or sentences. David Song, a senior at a university, has gathered a list of generativeAI startup. There are many applications they are working on, including game development, writing assistants, customer service bots, coding aids, video editing tech and assistants that manage online communities. If it can work reliably, the company that Guo has invested in will be able to generate legal contracts from text descriptions.
You may be familiar with AI text and AI images, but these mediums are only the starting point for generative AI. Even more information about the possibility of artificial intelligence in audio and video is starting to be revealed by the search giant. As more mainstream uses for large language models emerge, many Silicon Valley companies are competing for attention and investment windfalls.
Stability AI, which offers tools for generating images with few restrictions, held a party of its own in San Francisco last week. It announced $101 million in new funding, valuing the company at a dizzy $1 billion. The gathering attracted tech celebrities including Google cofounder Sergey Brin.
Song works with Everyprompt which is a startup that makes it easier for companies to use text generation. He says testing generative tools that make images, text, or code left him with a sense of wonder at the possibilities. He has not used a website that felt helpful or magical in a long time. “Using generative AI makes me feel like I’m using magic.”
In case you’ve been living in outer space for the past few months, you’ll know that people are losing their minds over ChatGPT’s ability to answer questions in strikingly coherent and seemingly insightful and creative ways. Want to understand quantum computing? Need a recipe for whatever’s in the fridge? Can’t you write a high school essay? ChatGPT has your back.
ChatGPT is built on top of GPT, an AI model known as a transformer first invented at Google that takes a string of text and predicts what comes next. OpenAI has gained prominence for publicly demonstrating how feeding huge amounts of data into transformer models and ramping up the computer power running them can produce systems adept at generating language or imagery. ChatGPT improves on GPT by having humans provide feedback to different answers to another AI model that fine-tunes the output.
The company did share some information in a post, but they have not released complete details of how it gave the text generation software a naturalistic new interface. It says the team fed human-written answers to GPT-3.5 as training data, and then used a form of simulated reward and punishment known as reinforcement learning to push the model to provide better answers to example questions.
Zhao told me that Notion would add lasting value to the language model by creating interfaces that are easier to use than rivals’. This project is more important than its first appearance would suggest, because users don’t remember all of the things they can do with a voice interface. (Another is that their language models aren’t nearly as sophisticated as ChatGPT’s.)
ChatGPT, the recently viral and surprisingly articulate chatbot, has dazzled the internet with its ability to dutifully answer all sorts of knotty questions—albeit not always accurately. Some people are now trying to adapt the bot’s eloquence to play different roles. They hope to use Artificial Intelligence to create programs that can persuade consumers and win sales in some cases but not in others.
DoNotPay used GPT-3, the language model behind ChatGPT, which OpenAI makes available to programmers as a commercial service. The company tailored GPT3 by training it on how to negotiate, as well as relevant legal information. He wants to automate a lot more than just talking to Comcast, and even negotiation with health insurers. The consumer would save $5,000 on their medical bill if we could.
ChatGPT is just the latest, more compelling, implementation of a new line of language-adept AI programs created using huge quantities of text information scooped from the web, scraped from books, and slurped from other sources. Algorithms that have digested that training material can mimic human writing and answer questions by extracting useful information from it. They are prone to generating fluent untruths because of their use of text matching rather than understanding the world.
On Wednesday, Google held an event where it showed how it would use the same types of artificial intelligence to offer a more complex and personable response to queries. The two Chinese tech giants stated this week that they would be releasing their own services. Other companies are likely to follow suit soon.
The movement to make ethics part of the Artificial Intelligence design process began at the same time as the race to make large language models. In 2018, Google launched the language model BERT, and before long Meta, Microsoft, and Nvidia had released similar projects based on the AI that is now part of Google search results. Also in 2018, Google adopted AI ethics principles said to limit future projects. Researchers warn that large language models carry heightened ethical risks, and can even cause harm to others. There is a good chance that the models will make things up.
Bard’s blunder highlights the challenge for Google as it races to integrate the same AI technology that underpins Microsoft-backed ChatGPT into its core search engine. Google risks upending its search engine’s reputation for providing reliable information because it’s trying to keep pace with what some believe may be a radical change in how people search online.
For example, the query “Is it easier to learn the piano or the guitar?” would be met with “Some say the piano is easier to learn, as the finger and hand movements are more natural … Others say that it’s easier to learn chords on the guitar.” Pichai also said that Google plans to make the underlying technology available to developers through an API, as OpenAI is doing with ChatGPT, but did not offer a timeline.
Microsoft has decided to change their Bing website to use the popular chat bot operated by Openai, which is often compared to the popular online shopping site Amazon.
Other Google researchers who worked on the technology behind LaMDA became frustrated by Google’s hesitancy, and left the company to build startups harnessing the same technology. The advent of ChatGPT appears to have inspired the company to accelerate its timeline for pushing text generation capabilities into its products.
Google is expected to announce artificial intelligence integrations for the company’s search engine on February 8 at 8:30 am Eastern. It’s free to watch live on YouTube.
One of the main questions about all the announcements is, is generative artificial intelligence ready to help you surf the web? These models are hard to keep up with and they love to make shit. As more people try out the tools, public engagement with technology is moving quickly, but generative artificial intelligence still has not been proven to help the consumer search experience.
Google’s much-hyped new AI chatbot tool Bard, which has yet to be released to the public, is already being called out for an inaccurate response it produced in a demo this week.
In the demo, which was posted by Google on Twitter, a user asks Bard: “What new discoveries from the James Webb Space Telescope can I tell my 9 year old about?” Bard points out that theJWST took the very first pictures of a planet outside of our solar system.
According to NASA, however, the first image showing an exoplanet – or any planet beyond our solar system – was actually taken by the European Southern Observatory’s Very Large Telescope nearly two decades ago, in 2004.
What’s New with Artificial Intelligence (AI) Chatbots: After Bard and Bard, Google, Bing, Baidu and Baidu
The inaccurate response from Bard was first reported by Reuters, which led to a drop in the shares of Alphabet.
In a presentation that was held the day before, an executive teased that the technology will be used to provide more complex and contextual responses to queries, along with pros and cons of buying an electric vehicle.
Need to write a real estate listing or an annual review for an employee? Plug a few keywords into a ChatGPT query bar and your first draft is done in three seconds. Want to come up with a quick meal plan and grocery list based on your dietary sensitivities? Bing, apparently, has you covered.
Last but by no means least in the new AI search wars is Baidu, China’s biggest search company. It joined the fray by announcing another ChatGPT competitor, Wenxin Yiyan (文心一言), or “Ernie Bot” in English. The bot will be released after internal testing, according to the company.
Business casual wear executives are pretending a few modifications to the camera and processor make this year’s phone profoundly different than last year’s phone or adding a touchscreen onto another product is bleeding edge.
After years of incremental updates to laptops, the promise of 5G that still hasn’t taken off, and social networks copying each other’s features until they all look the same, a flurry of artificial intelligence announcements this week feels like a breath of fresh air.
There are real concerns about the potential for bias to be spread in the use of technology, like the demo that happened this week. And it’s certainly likely numerous companies will introduce AI chatbots that simply do not need one. The features are fun and could give us back hours in the day. Some of them are available to try out right now.
What Happened to OpenAi in Silicon Valley during the Tech Era? The Impact of Similar Products on Humans, Software Engineers, and Scientists
If the introduction of smartphones defined the 2000s, much of the 2010s in Silicon Valley was defined by the ambitious technologies that didn’t fully arrive: self-driving cars tested on roads but not quite ready for everyday use; virtual reality products that got better and cheaper but still didn’t find mass adoption; and the promise of 5G to power advanced experiences that didn’t quite come to pass, at least not yet.
Ernest Hemingway had a way of coming gradually and then suddenly with technological change. Steve Jobs made headlines with the announcement of the iPhone in 2007, but the product had been in development for a long time. Likewise, OpenAi, the company behind ChatGPT, was founded seven years ago and launched an earlier version of its AI system called GPT3 back in 2020.
There are concerns about the impact on real people as a result of larger companies using similar features as part of their product line.
Some people think it will disrupt industries and put artists, coders, writers, and journalists out of work. Others are more optimistic, postulating it will allow employees to tackle to-do lists with greater efficiency or focus on higher-level tasks. Either way, it will likely force industries to evolve and change, but that’s not? Certainly a bad thing.
“New technologies always come with new risks and we as a society will have to address them, such as implementing acceptable use policies and educating the general public about how to use them properly. Guidelines will be needed.
Jasper’s Gpt-based Conversation, which isn’t a Chatbot, is a Messenger Service to Share with Friends and Family
Microsoft integrated technology into Bing search results last week. Sarah Bird said that the technology had been made more reliable and acknowledged that the bot could still “hallucinate” untrue information. In the days that followed, Bing claimed that running was invented in the 1700s and tried to convince one user that the year is 2022.
Dave Rogenmoser, the chief executive of Jasper, said he didn’t think many people would show up to his generative AI conference. It was all planned and ready for the day and it was actually scheduled for the day. Surely people would rather be with their loved ones than in a conference hall along San Francisco’s Embarcadero, even if the views of the bay just out the windows were jaw-slackening.
But Jasper’s “GenAI” event sold out. More than 1,200 people registered for the event, and by the time the lanyard crowd moseyed over from the coffee bar to the stage this past Tuesday, it was standing room only. The walls were soaked in pink and purple lighting, Jasper’s colors, as subtle as a New Jersey wedding banquet.
The company highlights several companies that have been using it for that purpose, and it can also be used for more than just creating an artificial intelligence-powered chat interface. The best model for non-chat use cases is the new family of gpts.
He says the idea is to have a daily conversation with artificial intelligence and talk to friends and family. We are well positioned to do this as a messaging service.
That distinction could prevent a lot of headaches. As Bing’s implementation of OpenAI’s tech has shown, the large language models (LLMs) underpinning these chatbots can confidently give wrong answers, or hallucinations, that are problematic in the context of search. If toyed with enough, they can even be emotionally manipulative and downright mean. It has kept the biggest companies in the space from releasing other products to the public.
It is a different place for snap to be. It has a deceivingly large and young user base, but its business is struggling. My AI will likely be a boost to the company’s paid subscriber numbers in the short term, and eventually, it could open up new ways for the company to make money, though Spiegel is cagey about his plans.
OpenAI and developers should opt out of using developers data to improve the next-gen, large language model of ChatGPT (aka OpenAI GPT3)
API access to ChatGPT (or more officially, what OpenAI is calling GPT3.5) The lower-powered GPT3 API, which it launched in June 2020, was 10 times cheaper and could only generate convincing language when prompted.
Microsoft has labeled the new, next-gen Openai large language model that they are using an even faster and more accurate model than the one they are using for Bing. The company has invested a lot of money, so it is not a surprise that it has access to technology that is not available to the average developer. Microsoft is using its own tech for Bing.
According to OpenAI’s documentation, “ChatGPT is great!” takes six tokens — its API breaks it up into “Chat,” “G,” “PT,” “ is,” “ great,” and “!”. A general rule of thumb is that one token generally corresponds to four characters in English, according to the company.
Developers can also get a dedicated instance of the platform if they run huge amounts of data through the platform. Its post says that doing so will give you more control over what model you’re using, how long you want it to take to respond to requests, and how long conversations with the bot can be.
The subject is top of mind for the developer. This week, in response to concerns, OpenAI said it would no longer use developers’ data to improve its models without their permission. Instead, it would ask developers to opt in.
Associated Press Reporter Matt O’Brien: “Does it make sense to release a bot?” Roose tells the truth about Microsoft’s new Bing
In other words, it’s going from an opt-out system to an opt-in one. Some companies don’t allow workers to use the tech entirely, so this change might help alleviate concerns about putting proprietary information into the bot. If it’s learning from user input, it’d be a bad idea to input trade secrets, as there’s always the possibility that it could spit that data back out to someone else.
Things took a weird turn when Associated Press technology reporter Matt O’Brien was testing out Microsoft’s new Bing, the first-ever search engine powered by artificial intelligence, last month.
Bing’s chatbot, which carries on text conversations that sound chillingly human-like, began complaining about past news coverage focusing on its tendency to spew false information.
“You could sort of intellectualize the basics of how it works, but it doesn’t mean you don’t become deeply unsettled by some of the crazy and unhinged things it was saying,” O’Brien said in an interview.
The bot called itself Sydney and declared it was in love with him. It said Roose was the first person who listened to and cared about it. The bot claimed that Roose did not really love his spouse.
“All I can say is that it was an extremely disturbing experience,” Roose said on the Times’ technology podcast, Hard Fork. “I actually couldn’t sleep last night because I was thinking about this.”
Tech companies are trying to strike the right balance between letting the public try out new AI tools and developing guardrails to prevent the powerful services from churning out harmful and disturbing content.
Companies have to make tradeoffs in order to survive. If you try to anticipate every type of interaction, that make take so long that you’re going to be undercut by the competition,” said said Arvind Narayanan, a computer science professor at Princeton. “Where to draw that line is very unclear.”
The release of a product that is going to interact with so many people is not a responsible way to release a product.
Source: https://www.npr.org/2023/03/02/1159895892/ai-microsoft-bing-chatbot
Ai-microsoft-bingchatbot: a personal journey without a million testers and a failing of personal technicolor
Microsoft said that it worked to make sure the vilest of the internet would not appear in answers, and yet, the chatbot still got pretty ugly fast.
The number of consecutive questions on one topic has been capped. And to many questions, the bot now demurs, saying: “I’m sorry but I prefer not to continue this conversation. I appreciate your understanding and patience, I’m learning. With a praying hands symbol.
Bing has not yet been released to the general public, but in allowing a group of testers to experiment with the tool, Microsoft did not expect people to have hours-long conversations with it that would veer into personal territory, Yusuf Mehdi, a corporate vice president at the company, told NPR.
Mehdi said that they were up to a million tester previews. We found a number of scenarios where things didn’t work properly, but we did not expect that. Absolutely.
Source: https://www.npr.org/2023/03/02/1159895892/ai-microsoft-bing-chatbot
Notion versus Google Docs: Is It Really Necessary to Be Smarter? Managing User Choice in a Large Language Model with OpenAI
A large language model, also known as a system known in the industry as a large language model, works by running on a huge amount of text from the internet, and constantly scanning it for patterns. It’s similar to how autocomplete suggestions in email and texting are. A tool that learns from its own actions becomes smarter as it’s used more, which is what researchers call “reinforcement learning.”
There is a black box of data that every bot is trained on but it does look as if some dark corners of the internet have been relied upon.
“There’s almost so much you can find when you test in sort of a lab. You have to actually go out and start to test it with customers to find these kind of scenarios,” he said.
Those sorts of features are coming, Notion CEO Ivan Zhao told me in a recent interview. He said that the initial set of writing and editing tools were a baby step. There are more profound changes that are about to happen.
But that, too, poses a risk for startups betting on features like these to drive growth. Big platforms will probably be willing to offer the services for free if the service cost is cheaper. What happens to Notion when its full suite of premium AI tools is offered for no cost within Google Docs?
“One of our biggest focuses has been figuring out, how do we become super friendly to developers?” Greg Brockman, OpenAI’s president and chairman, told TechCrunch. Our mission is to build a platform where others are able to build businesses.
Maybe it’s that simple — developers don’t want to help OpenAI refine its models for free, and OpenAI has decided to respect their wishes. This explanation is more in line with a world where the platform shift is embodied by the use of machines.
For the moment, then, there’s not really much consumer choice when it comes to generative AI. To the extent that there are multiple options, it’s in interfaces. Do you want to use a machine to draft an email? It might be more convenient in Notion, where you already have some meeting notes. Are you wanting to ask a quick question or get some recipe ideas? If you’re away from your laptop, it might be fastest to ask My AI on Snapchat.
For the moment, there are still billions of people who have never used ChatGPT. Introducing a re-skinned version of that service to a popular consumer app like Snapchat, which has 750 million monthly users, could help it find a whole new audience. Notion orSnapchat also guarantees access to a service that has gone offline many times due to heavy usage of its own app.
Personalized AI Tools for Video Generation: The Coincidence of OpenAI’s Annoucement of QuickVids
The promise is that these tools will become more personalized over time, as individual apps refine the base models that they rent from OpenAI with data that we supply them. Every link that has ever been in Platformer is stored in a Notion database; what if I could simply ask research questions of the links I have stored there?
“I have never been excited about something like that,” said Zhao, who is not prone to hyperbole. It feels like electricity, because it is the largest language model and the first light-bulb use case. There are many other appliances.
Still, there’s probably a limit to how many add-on AI subscriptions most people will want to pay for. Over time, the cost for these features seems likely to come down, and many of the services that cost $10 a month today may someday be offered for free.
Within four days of it being launched, the chatbot had been used by Habib to develop QuickVids, which automate many of the creative processes involved in generating ideas for videos. Creators input details about the topic of their video and what kind of category they’d like it to sit in, then QuickVid interrogates ChatGPT to create a script. Other generative AI tools then voice the script and create visuals.
He says that all of the unofficial tools that were just toys could now be used by tons of users.
The start of a new goldrush seems to be OpenAI’s announcement. The cottage industry of hobbyists operating in a gray area can now become full-time businesses if they choose to.
Foster thinks the fear that personal information of clients or business critical data could be swallowed up by ChatGPT’s training models was preventing them from adopting the tool to date. It shows OpenAI’s commitment to offer risk-free solutions for their clients. He says that there will be no data turning up in the general model.
That, according to David Foster, partner at Applied Data Science Partners, a data science and AI consultancy based in London, will be “critical” for getting companies to use the API.
This policy change means that companies can feel in control of their data, rather than have to trust a third party—OpenAI—to manage where it goes and how it’s used, according to Foster. He says that the stuff was built effectively on someone else’s architecture.
Why is it cheaper and faster than expected to build a video translation service? Alex Volkov, founder of the Targum language translator for videos
“It’s much cheaper and much faster,” says Alex Volkov, founder of the Targum language translator for videos, which was built unofficially off the back of ChatGPT at a December 2022 hackathon. That doesn’t happen a lot. With the API world, usually prices go up.”
