Google Bard, OpenAI ChatGPT, Bing Highlights: Alphabet shares dive due to Bard’s incorrect answer

Bing uses a more advanced version of the LLM (large language model) that ChatGPT employs, and as such, offers chatbot features that help you generate text. In the video below, you can see Bing generate a poem based on a short prompt.

Demos for the new AI-powered Bing are starting to show up on Twitter and YouTube. Watch how the one shared below understands and summarises a 15-page PDF in mere seconds. The function should help users analyse lenghty raw data more and manage their workflows more efficiently.

This is the most impressive feature of the new bing.

The GPT browser can understand and summarize a 15-page PDF in seconds.

You can now ask for the the key takeaways of each page and chat about the content of the document. pic.twitter.com/OSa4GztdAA

— Lior? (@AlphaSignalAI) February 10, 2023

Yes chatbots like ChatGPT seem perfect and designed to replace humans. But as our Nandagopal Rajan writes in this opinion piece, that won't happen just yet, as these bots are far from perfect. "The new buzz around AI is mostly thanks to OpenAI’s ChatGPT chatbot which can handle most queries humans throw at it. But what it does is not really new. It has just become better at its job. And even then it still cannot match what human intelligence is capable of. For example, if you ask the bot about a certain person, with a very common name, the chances are it will give you a response mixing the profiles of multiple people with the same name. That’s because it still has a lot to learn." Read the full opinion piece here.

According to Reuters, which first spotted the incorrect answer by Bard, Google has given a response as well.  "This highlights the importance of a rigorous testing process, something that we're kicking off this week with our Trusted Tester program," a Google spokesperson said. "We'll combine external feedback with our own internal testing to make sure Bard's responses meet a high bar for quality, safety and groundedness in real-world information."  

On Google competing with Bing’s new AI, Microsoft CEO Satya Nadella said in a The Verge interview, “I hope with our innovation they will definitely want to come out and show that they can dance. I want people to know that we made them dance."

For now, AI language models can change how search is done, and affect the tech industry. But unlike its predecessors, the software can now learn and improve without human intervention.

From Google's Bard to Microsoft's AI-powered Bing search engine, several big tech companies have joined the AI race. If you want to learn about the key developments in the field of artificial intelligence, click here.

Bard has no doubt got off to a rough start thanks to that incorrect answer around the James Webb Space Telescope. Google is so far only opened this up to 'trusted testers'. We're guessing public testing will take longer. 

With Microsoft Bing getting a chat-like interface powered by Open AI's language models, search is bound to change in a big way, writes Anuj Bhatia. And for Google this will cause some serious trouble. Search could no longer mean a number of links, but an actual coherent answer. Read more here.

Well, it looks like an incorrect answer given by Google Bard, which was shared via a tweet, has caused chaos of the company. Alphabet -- Google parent-- saw its shares take a sharp dive, their biggest fall in the last three months. Google had shared a screenshot of the same on Twitter, and it looks like this answer has caused chaos. In the answer, Bard claims that the James Webb Space Telescope was one of the first to take the first pictures of a planet outside the Earth’s solar system, of an 'exo-planet'. But NASA has said the picture were taken by a different telescope. The end result: Google has lost over $100 billion in market valuation. 

Google's AI event is now over. We didn't get any new information. It mostly Google recounting everything it has done in AI so far. And we know that Bard is still going to be in 'limited testing' for now. But Search could soon have more in-depth, essay-like answers. Exactly when though? Google hasn't specified a timeline. 

The 'Blob Opera' AI experiment is one where a neural network or AI creates what opera might sound like. The project was a machine learning experiment by David Li in collaboration with Google Arts & Culture. 

It is expanding access for Indoor Live View to several cities and 1000 new places. It will also roll out more widely on Android and iOS. 

Yes, Google Maps will get some more AR features driven by AI as well. You could soon point the camera on your phone to a coffee shop, etc and get to see Google reviews about the place shown on the screen. But, of course, this is not coming to India. 

Google now talking about the Immersive view on Maps which relies on AI as well. Keep in mind, this feature was showcased earlier as well, and it has not rolled out in all places. Google will start rolling it out today in Los Angeles, London, etc. Immersive View was originally showcased back in September.

Raghavan says Google will continue to focus on 'responsible' AI. He says the company wants to deliver AI in way that is bold and responsible. 

Yes, Search is changing as Pichai wrote in the blog post. Soon you will be able to ask complex questions and get detailed answers--in the form of an essay-- as part of Search Results as well. These answers are being created using 'generative AI', says Google. It has said it will try and ensure a balance that these responses do not impact traffic on websites, which is a fair and serious concern. 

Google is now giving a detailed demo of Bard, which is based on LaMDA. But the demo was really short. Also Google is limited Bard's testing to just select 'trusted testers'. So this is not open for public testing yet. It looks like Google wants to make sure 'Bard' causes the least amount of trouble before it is made available for public testing. 

Finally, Google is talking about Bard. 

Google says Multisearch will soon work with any image you see. This will allow users to get more information and context based on the image itself. 

The Search Screen feature will let users search for any information or video or image they might have open on any other app. So if a friend sends you a video with an interesting artwork in it, you will be able to open it, call up Google Assistant, and ask it to Search for information about the video. Google is also talking about the multi-search feature, which it has showcased earlier. 

Raghavan is now talking about Google Lens, which is used more than 10 billion times a month.  It is no longer a novelty, says Google. Lens' translate feature for text is now rolling out globally on Android. Google had announced this feature last September. 

Google says over 1 billion people are using its translate feature. Raghavan is talking about how the translate feature was used extensively by Ukrainians when they were looking for refuge in other countries. Google says it will bring a more richer version to translate as well which will give more context when one is translating words. Google uses Zero Shot Machine Translation for translation, and it has now added 24 more languages to translation with the help of this model. 

Google SVP Prabhakar Raghavan on stage. He's starting by talking about how search now has a deeper understanding of language and context. He's talking how many of Google Search's features are now driven by AI-- such as Google Lens, AR in search or how AR directions show up in Google Maps in select cities. 

In a recent blog post, Google and Alphabet CEO Sundar Pichai had made it clear that AI-powered features are coming to search as well. While not all of us will get to try out Bard immeditately, it looks like Google search will soon showcase "complex information and multiple perspectives into easy-to-digest formats," aka the essay format that we've seen ChatGPT executve so well. Google says these formats will help users get a deeper perspective around a query. How soon we will see these changes to search? Well that the time is not yet specified. 

The livestream is on. We have another 10 minutes or so to go once the keynote begins. 

LaMDA stands of Language Model for Dialogue Applications or basically a language model that can have a conversation with us-- just like another human. Google showcased LaMDA back in 2021 at its annual I/O developer conference. But LaMDA has been in a limited test till date. Users have to get accepted to the AI Test Kitchen app's waitlist in order to try out LaMDA. When it was revealed, Google had revealed that LaMDA was trained using "human dialogue and stories," and could engage in conversations. Google also recently showed how LaMDA is being used to help authors write works of fiction. Read more on that here.

According to the Google event, they plan to showcase how they are making it more "natural and intuitive than ever before to find what you need." So yes, we should see more features related to AI and search. Google's description also mentions Maps and beyond, so we could also see some AI-specific features coming to products beyond Search as well. 

Well Bard is not yet open to all. It looks like Google is taking a cautious approach to its AI-- just like it did with LaMDA. Bard will only made available as part of beta testing, but people will have to sign up for this. Hopefully we will get to know more details on where to sign up for this chatbot. 

Google is playing catch up give Microsoft has fired some serious shots as it showcased plans to integrate Bing with AI-features driven by the same technology that is powering ChatGPT as well. The OpenAi and Microsoft partnership is a big threat to Google's search business and the company will need to showcase how it will tackle this threat. Today's AI event from Paris will hopefully give some clear indicators. 

The most anticipated announcement during Google’s AI event will be the launch of Bard, the chatbot aimed at countering the popularity of ChatGPT. According to a blog post by Sunder Pichai, Bard will initially be available only to a group of “trusted testers” before it gets a wider release later this year. But what exactly should we expect from it? 

For the most part, it would probably come with many of the same functionalities as ChatGPT. According to AP, Bard will be able to explain complex subjects like outer space discoveries in simple enough terms for a child to understand. The service should also be capable of more mundane tasks like providing party planning trips or lunch recipe ideas.

But in his blog post, Pichai did not mention whether Bard will be able to pull off some of ChatGPT’s party tricks like writing in the style of William Shakespeare or Oscar Wilde. But Pichai did write that “Bard can be an outlet for creativity, and a launchpad for curiosity” in the blog post. 

Since chatbots like ChatGPT learn by analysing huge amounts of data posted to the internet, there is a chance that they will internalise some of the biases that are contained within that data. According to NYT, this means that chatbots could end up generating toxic language and hate speech. They can also deliver information that is biased against women and people of colour. 

This fact works against Google because, unlike newer companies like OpenAI, Google cannot risk damaging the corporate brand that it has built over decades. To tackle these and other issues, teams from Google research, Trust & Safety and other departments have been reassigned to help develop and release new AI products, according to an NYT report.

LaMDA or Language Models for Dialog Applications is a machine-learning language model is a machine learning model that mimics humans in conversation. Like BERT and GPT (which underpins ChatGPT), LaMDA is built on Transformer, a neural network architecture that Google built and open-sourced in 2017. One impressive aspect of LaMDA is its ability to conduct fluid, open-ended conversations, where it can start at one topic and end up at another, traversing different topics and subjects. 

The most immediate threat to Google’s search business from ChatGPT will be with the latter’s integration with Microsoft’s Bing search engine. For a long time, Microsoft’s underdog search engine has remained a distant second to the search engine behemoth’s offering. But all that could change if Bing starts giving answers in the conversational style that ChatGPT is known for.

OpenAI made waves when it revealed ChatGPT and its ability to package information in clear and simple sentences instead of serving up a bunch of relevant links. Combine this with its ability to explain concepts simply and the way in which it can seemingly generate ideas from scratch, and the alarm bells had already begun ringing for Google.

Even though ChatGPT clearly has a lot to improve, its release prompted Google’s management to declare a “code red,” according to The New York Times. This is because ChatGPT represents an enormous technological change that can upend Google’s search and advertising business. 

In fact, Google pulled out the big guns in its fight against this seemingly existential threat—it called in founders Larry Page and Sergey Brin. According to NYT, Page and Brin haven’t spent much time with Google since they left their daily roles with the company in 2019. The pair reviewed Google’s AI product strategy and also approved plans and pitched ideas to put more chatbot features in Google’s search engine. 

One of the main expectations from Google’s AI event today is that the company will reveal its Bard chatbot. Staff at Google have been testing the in-house AI chatbot that responds to user prompts in a way similar to OpenAI’s ChatGPT. It is based on LaMDA (or Language Model for Dialogue Applications), the language model that became famous when a Google engineer thought it had gone sentient. 

Based on reports, Bard will hold an important advantage over ChatGPT—the former can integrate recent events into responses. ChatGPT,  on the other hand, has limited knowledge of events after 2021. 

This was evident when according to CNBC, one of the testers asked Bard if there would be another round of layoffs at Google; the chatbot responded by saying, “According to my database access, it is unlikely for Google to conduct another round of layoffs in 2023… Layoffs are generally conducted to reduce costs and structure, but the company is doing well financially. In fact, Google’s revenue increased by 34% in 2021, and the company’s stock price has risen by 70% since January 2022.