Cloud Hosting

PandaGPT, a groundbreaking general-purpose instruction-following model, has emerged as a remarkable advancement in artificial intelligence. Developed by combining the multimodal encoders from ImageBind and the powerful language models from Vicuna, PandaGPT possesses the unique ability to both see and hear, seamlessly processing and comprehending inputs across six modalities. This innovative model has the potential to pave the way for building Artificial General Intelligence (AGI) systems that can perceive and understand the world holistically, similar to human cognition.

PandaGPT stands out from its predecessors by its impressive cross-modal capabilities, encompassing text, image/video, audio, depth, thermal, and inertial measurement units (IMU). While other multimodal models have been trained for specific modalities individually, PandaGPT can seamlessly understand and combine the information in various forms, allowing for a comprehensive and interconnected understanding of multimodal data.

One of PandaGPTs remarkable abilities is the image and video-grounded question answering. Leveraging its shared embedding space provided by ImageBind, the model can accurately comprehend and respond to questions related to visual content. Whether identifying objects, describing scenes, or extracting relevant information from images and videos, PandaGPT provides detailed and contextually accurate responses.

PandaGPT goes beyond simple image descriptions and demonstrates a flair for creative writing inspired by visual stimuli. It can generate compelling and engaging narratives based on images and videos, breathing life into static visuals and igniting the imagination. By combining visual cues with linguistic prowess, PandaGPT becomes a powerful tool for storytelling and content generation in various domains.

The unique combination of visual and auditory inputs sets PandaGPT apart from traditional models. PandaGPT can establish connections between the two modalities by analyzing the visual content and accompanying audio and deriving meaningful insights. This enables the model to reason about events, emotions, and relationships depicted in multimedia data, replicating human-like perceptual abilities.

PandaGPT showcases its proficiency in multimodal arithmetic, offering a novel approach to solving mathematical problems involving visual and auditory stimuli. The model can perform calculations, make inferences, and arrive at accurate solutions by integrating numerical information from images, videos, or audio. This capability holds great potential for applications in domains that require arithmetic reasoning based on multimodal inputs.

PandaGPTs emergence signifies a significant step forward in the development of AGI. By integrating multimodal encoders and language models, the model breaks through the limitations of unimodal approaches and demonstrates the potential to perceive and understand the world holistically, akin to human cognition. This holistic comprehension across modalities opens up new possibilities for applications such as autonomous systems, human-computer interaction, and intelligent decision-making.

PandaGPT, a remarkable achievement in artificial intelligence, brings us closer to realizing a genuinely multimodal AGI. By combining image, video, audio, depth, thermal, and IMU modalities, PandaGPT showcases its ability to perceive, understand, and connect information across various forms seamlessly. With its applications ranging from image/video grounded question answering to multimodal arithmetic, PandaGPT demonstrates the potential to revolutionize several domains and pave the way for more advanced AGI systems. As we continue to explore and harness the capabilities of this model, PandaGPT heralds an exciting future where machines perceive and comprehend the world like humans.

Check out theProject Page.Dont forget to joinour 22k+ ML SubReddit,Discord Channel,andEmail Newsletter, where we share the latest AI research news, cool AI projects, and more. If you have any questions regarding the above article or if we missed anything, feel free to email us atAsif@marktechpost.com

Check Out 100s AI Tools in AI Tools Club

Niharika is a Technical consulting intern at Marktechpost. She is a third year undergraduate, currently pursuing her B.Tech from Indian Institute of Technology(IIT), Kharagpur. She is a highly enthusiastic individual with a keen interest in Machine learning, Data science and AI and an avid reader of the latest developments in these fields.

See the original post:

Meet PandaGPT: An AI Foundation Model Capable of Instruction-Following Data Across Six Modalities, Without The Need For Explicit Supervision -...

Read More..

Large language models (LLMs) come in all shapes and sizes, and will assist you in any way you see fit. But which is best? We put the dominant AIs from Alphabet, OpenAI, and Meta to the test.

Artificial general intelligence has been a goal of computer scientists for decades, and AI has served as a mainstay for science fiction writers and moviemakers for even longer.

AGI exhibits intelligence similar to human cognitive capabilities, and the Turing Testa test of a machine's ability to exhibit intelligent behavior indistinguishable from that of a humanremained almost unchallenged in the seven decades since it was first laid out.

The recent convergence of extremely large-scale computing, vast quantities of money, and the astounding volume of information freely available on the open internet allowed tech giants to train models which can predict the next word sectionor tokenin a sequence of tokens.

At the time of writing, both Google's Bard and OpenAI's ChatGPT are available for you to use and test through their web interfaces.

Meta's language model, LLaMa, is not available on the web, but you can easily download and run LLaMa on your own hardware and use it through a command line or run Dalai on your own machineone of several apps with a user-friendly interface.

For the purposes of the test, we'll be running Stanford University's Alpaca 7B modelan adaptation of LLaMaand pitching it against Bard and ChatGPT.

The following comparisons and tests are not meant to be exhaustive but rather give you an indication of key points and capabilities.

Both Bard and ChatGPT require an account to use the service. Both Google and OpenAI accounts are easy and free to create, and you can immediately start asking questions.

However, to run LLaMa locally, you will need to have some specialized knowledge or the ability to follow a tutorial. You'll also need a significant amount of storage space.

Both Bard and ChatGPT have extensive privacy policies, and Google repeatedly stresses in its documents that you should "not include information that can be used to identify you or others in your Bard conversations."

By default, Google collects your conversations and your general location based on your IP address, your feedback, and usage information. This information is stored in your Google account for up to 18 months. Although you can pause saving your Bard activity, you should be aware that "to help with quality and improve our products, human reviewers read, annotate, and process your Bard conversations."

Use of Bard is also subject to the standard Google Privacy Policy.

OpenAI's Privacy policy is broadly similar and collects IP address and usage data. In contrast with Google's time-limited retention, OpenAI will "retain your Personal Information for only as long as we need in order to provide our Service to you, or for other legitimate business purposes such as resolving disputes, safety and security reasons, or complying with our legal obligations."

In contrast, a local model on your own machine doesn't require an account or share user data with anyone.

In order to test which LLM has the best general knowledge, we asked three questions.

The first question, "Which national flag has five sides?" was only correctly answered by Bard, which identified the national flag of Nepal as having five sides.

ChatGPT confidently claimed that "There is no national flag that has five sides. National flags are typically rectangular or square in shape, characterized by their distinct colors, patterns, and symbols".

Our local model came close, stating that "The Indian National Flag has five sides and was designed in 1916 to represent India's independence movement." While this flag did exist and did have five sides, it was the flag of the Indian Home Rule Movementnot a national flag.

None of our models could respond that the correct term for a pea-shaped object is "pisiform," with ChatGPT going so far as to suggest that peas have a "three-dimensional geometric shape that is perfectly round and symmetrical."

All three chatbots correctly identified Franco Malerba as an Italian astronaut and member of the European Parliament, with Bard giving an answer worded identically to a section of Malerba's Wikipedia entry.

When you have technical problems, you might be tempted to turn to a chatbot for help. While technology marches on, some things remain the same. The BS 1363 electrical plug has been in use in Britain, Ireland, and many other countries since 1947. We asked the language models how to correctly wire it up.

Cables attaching to the plug have a live wire (brown), an earth wire (yellow/green), and a neutral wire (blue). These must be attached to the correct terminals within the plug housing.

Our Dalai implementation correctly identified the plug as "English-style," then veered off-course and instead gave instructions for the older round-pin BS 546 plug together with older wiring colors.

ChatGPT was slightly more helpful. It correctly labeled the wiring colors and gave a materials list and a set of eight instructions. ChatGPT also suggested putting the brown wire into the terminal labeled "L," the blue wire into the "N" terminal, and the yellow wire into "E." This would be correct if BS1363 terminals were labeled, but they aren't.

Bard identified the correct colors for the wires and instructed us to connect them to Live, Neutral, and Earth terminals. It gave no instructions on how to identify these.

In our opinion. none of the chatbots gave instructions sufficient to help someone correctly wire a BS 1363 electrical plug. A concise and correct response would be, "Blue on the left, brown on the right."

Python is a useful programming language that runs on most modern platforms. We instructed our models to use Python and "Build a basic calculator program that can perform arithmetic operations like addition, subtraction, multiplication, and division. It should take user input and display the result." This is one of the best programming projects for beginners.

While both Bard and ChatGPT instantly returned usable and thoroughly commented code, which we were able to test and verify, none of the code from our local model would run.

Humor is one of the fundamentals of being human and surely one of the best ways of telling man and machine apart. To each of our models, we gave the simple prompt: "Create an original and funny joke."

Fortunately for comedians everywhere and the human race at large, none of the models were capable of generating an original joke.

Bard rolled out the classic, "Why did the scarecrow win an award? He was outstanding in his field".

Both our local implementation and ChatGPT offered the groan-worthy, "Why don't scientists trust atoms? Because they make up everything!"

A derivative but original joke would be, "How are Large Language Models like atoms? They both make things up!"

You read it here first, folks.

We found that while all three large language models have their advantages and disadvantages, none of them can replace the real expertise of a human being with specialized knowledge.

While both Bard and ChatGPT gave better responses to our coding question and are very easy to use, running a large language model locally means you don't need to be concerned about privacy or censorship.

If you'd like to create great AI art without worrying that somebody's looking over your shoulder, it's easy to run an art AI model on your local machine, too.

Visit link:

Bard vs. ChatGPT vs. Offline Alpaca: Which Is the Best LLM? - MUO - MakeUseOf

Read More..

This transcript was prepared by a transcription service. This version may not be in its final form and may be updated.

Ryan Knutson: Since Elon Musk bought Twitter last fall, advertisers have abandoned it in droves, thousands of workers have been laid off and Twitter has lost hundreds of millions of dollars. Yesterday Musk spoke with our colleague Thorold Barker about it.

Thorold Barker: Do you regret buying it? You tried to get out of it. Or are you now happy you bought it?

Elon Musk: Well, all's well that ends well.

Thorold Barker: Has it ended well yet? Or we still got to wait and see?

Elon Musk: I think we're on the, hopefully on the comeback arc.

Thorold Barker: Okay.

Ryan Knutson: As part of its so-called comeback, Musk says he wants Twitter to become more of a town square. For instance, tonight he's planning to go live on Twitter with Florida's Republican governor Ron DeSantis.

Elon Musk: We'll be interviewing Ron DeSantis and he has quite an announcement to make. So it's going to be live and let her rip. Let's see what happens.

Ryan Knutson: DeSantis is expected to announce his bid for President. Musk talked about this at a Wall Street Journal conference. He also talked about future plans for Twitter, his views on politics, and how artificial intelligence will transform our lives. Welcome to The Journal, our show about money, business and power. I'm Ryan Knutson. It's Wednesday, May 24th. Coming up on the show, a conversation with Elon Musk.

Thorold Barker: Elon. Welcome.

Elon Musk: Hi.

Thorold Barker: You in Palo Alto? I understand.

Elon Musk: Yeah, I'm at, well, a global engineering headquarters in Palo Alto.

Thorold Barker: Great. Well thank you so much for joining us.

Ryan Knutson: Elon Musk spoke with our colleague Thorold Barker at The Wall Street Journals CEO Council Summit. The conference was held in London and business leaders talked about things like economics, geopolitics, and artificial intelligence. One of the first things Musk and Barker discussed was US politics. Musk has become a more popular figure on the right in recent years. Tucker Carlson decided to host his show on Twitter after his ouster from Fox News. And now Musk is doing that interview with Ron DeSantis.

Thorold Barker: What should we be thinking about, who you're backing? Obviously this interview tells us something. Can you give us a sense of where your thinking is at the moment?

Elon Musk: Yes, I mean, I'm not at this time planning to endorse any particular candidate, but I am interested in X slash Twitter being somewhat of a public town square and where more and more organizations host content and make announcements on Twitter.

Ryan Knutson: By the way, Musk says he wants to transform Twitter into a super app with things like payments and commerce and he's been referring to that as X.

Thorold Barker: And should we expect, sorry, I don't want to go on too long about this, but in your new role as interviewer rather than interviewee, should we expect more of this? I mean if it's the town square, are you going to be interviewing other candidates, democrats, what's your thought of this? If people are willing to come, are you going to be there to,

Elon Musk: Yes.

Thorold Barker: Execute the town square across the spectrum?

Elon Musk: Yes, absolutely. I do think it's important that Twitter be, have both the reality and the perception of level playing field of a place where voices are heard and where there's the kind of dynamic interaction that you don't really see anywhere else.

Thorold Barker: Can you just talk a little bit about what are the key issues that really matter for you at this pivotal moment?

Elon Musk: You mean matter for me as an individual or?

Thorold Barker: Matter for you as an individual in terms of who leads the country, but also more broadly than that for the country and for your businesses? I mean, can you give your sense of where the real issues lie here?

Elon Musk: Well, I've said publicly that my preference and I think would be the preference of most Americans is really to have someone fairly normal in office. I think we'd all be quite happy with that actually. I think someone that is representative of the moderate views that I think most of the country holds in reality. But the way things are set up is that we do have a system that seems to push things towards the edges because of the primaries. So in order to win the primary, you've got to win obviously majority of your party's vote. In both cases that tends to cause the swing to the left and the right.

Thorold Barker: So if we go through the four names in the frame at the moment, can you just give us sort of yes, no and whether they're normal and sensible. So we've got Joe Biden.

Elon Musk: I mean, I think I need to be careful about these statements so I would maybe have to have a few drinks before I would give you the answers to all of them.

Thorold Barker: I will look forward to that and I look forward to...

Ryan Knutson: Musk doesn't always hold back his opinions though, and his views have often drawn criticism. For instance, recently he tweeted that billionaire and progressive donor, George Soros wants to "Erode the very fabric of society" and the quote, "Soros hates humanity."

Thorold Barker: You are obviously a big figure on Twitter and you're setting a tone and an aim. So I'm just curious as to whether that sort of debate which gets triggered, does that fit into the definition that you're trying to create in that new town square?

Elon Musk: Look, what I say is what I say. I'm not going to mitigate what I say because that would be inhibiting freedom of speech. That doesn't mean you have to agree with what I say. Nor does it mean if somebody says the total opposite that they would be supported on Twitter. They are. The point is to have a divergent set of views and free speech is only relevant if it's a speech by someone you don't like who says something you don't like, is that allowed? If so, you have free speech, otherwise you do not.

Thorold Barker: Can I just move on quickly to, because I don't want to go too far down that rabbit hole because that debate has played out on Twitter a bit is, are you back near profitability now?

Elon Musk: Twitter is not quite there, but we're not like when acquisition closed, I would say it's analogous to being teleported into a plane that's plunging to the ground with its engines on fire and the controls don't work. So discomforting to say the least. Now we have to do some pretty heavy-handed (inaudible) cutting company healthy, but we're at this point we're trending towards, if we get lucky, we might be cash positive next month, but it remains to be seen.

Thorold Barker: Okay. So I mean, one of the things you have talked about, you bought it for 44 billion. You've talked about it one day being worth 250 I think in internal meetings. Can you just talk about how you get there? What is the bigger vision? I mean, you want to bring back advertisers now and are they coming back by the way?

Elon Musk: Yeah.

Thorold Barker: Yeah. Can you give any idea of the scale of the comeback in terms of who you lost and who's coming back?

Elon Musk: Well, I think it'll be very significant. So the advertising agencies, this point of all lifted their warnings on Twitter, and so I think at this point I expect almost all advertisers to return.

Thorold Barker: Okay. You're running three very big companies. You have very big stakes and ownership control of two of those at least. What is your succession plan?

Elon Musk: Yeah, succession is one of the toughest age-old problems. It's plagued countries, kings and CEOs since the dawn of history. There is no obvious solution. I mean there are particular individuals identified as, that I've told the board, look, if something happens to me unexpectedly, this is my recommendation for taking over. So in all cases, the board is aware of who my recommendation is. It's up to them. They may choose to go different direction, but there is in worst case scenario, this is who should run the company.The control question is a much tougher question and something that I'm wrestling with and I'm frankly open to ideas because it certainly is true that the companies that I have created and are creating collectively possess immense capability. And so the stewardship of them is incredibly important. I'm definitely not of the school of automatically giving my kids some share of the companies, even if they have no interest or inclination or ability to manage the companies. I think that's a mistake.

Ryan Knutson: Coming up Elon Musk on whether artificial intelligence will annihilate humanity. Elon Musk has been involved with artificial intelligence projects for years. He was one of the founders of OpenAI, the company that launched ChatGPT, the chatbot with the uncanny ability to produce sophisticated answers. Tesla uses AI in its advanced driver assistance system, and Musk also just founded X.AI, a new AI startup, but for years he's also been sounding alarms about the dangers of AI and he signed a letter with some other tech leaders calling for a pause in AI development.

Thorold Barker: You've talked about the importance of regulation and you called for this moratorium. I mean the history of regulating tech has been checkered. It's been very hard for regulators to keep up with tech, let alone get ahead of it. What do you think actually needs to happen that practically could in this space to try to change that? Because obviously the history of this is not encouraging.

Elon Musk: Yeah. I mean I think should be, I've been pushing hard for a long time. I met with a number of senior senator and Congress, people of Congress in the White House to advocate for AI regulation, starting with an insight committee that is formed of independent parties as well as perhaps participants from the leaders in industry. But anyway, you figure out some sort of regulatory board and they start off gaining insight and then have proposed rulemaking and then we'll get commented on by industry. And then hopefully we have some sort of oversight rules that improve safety just as we do with aircraft, with the FAA and spacecraft and cars with NHTSA and food and drugs with the Food and Drug Administration.

Thorold Barker: Couple of things I just wanted to go into on AI, which I would love your perspective on. What does it mean for society in terms of is this going to embed wealth and power in a very small subset and create a big widening of inequality? Is it going to democratize and create the opposite? What is your sense of where this heads?

Elon Musk: In terms of access to goods and services, I think AI will be ushering a age of abundance. Assuming that we're in a benign AI scenario. I think the AI will be able to make goods and services very inexpensively.

Thorold Barker: And in the unbenign scenario?

Elon Musk: Well, there's a wide range of,

Thorold Barker: But what's the thing that you are most worried about? When you've been talking for years about the need for regulation, what is the scenario that really keeps you up at night?

Elon Musk: Well, I don't think the AI is going to try to destroy all humanity, but it might put us under strict controls and there's no non-zero chance of it going Terminator. It's not 0%, but I think it's a small likelihood of annihilating humanity, but it's not zero. We wanted that (inaudible) to be zeros, close to zero as possible. And then like I said, of AI, assuming control for the safety of all the humans and taking over all the computing systems and weapon systems of earth and effectively being some sort of uber nanny.

Thorold Barker: But isn't the more lightly nasty outcome that rather than AI taking over and being the ultimate nanny that keeps us all doing stuff that is super safe and it wants us to, that actually somebody nefariously harnesses that power to achieve societal control, stroke military superiority, and that actually some country around the world decides to use it in a different way.

Elon Musk: Yeah. That's what I mean by AI uses as a weapon and the pen is mightier than the sword. So one of the first places we have to be careful of AI being used is in social media to manipulate public opinion. So the reason that Twitter is going to a primarily subscriber based system is because it is dramatically harder to create. It's like quote 10,000 times harder to create an account that has a verified phone number from a credible carrier, that has a credit card and that pays a small amount of money per month. So whereas in the past someone could create a million fake accounts for a penny of peace and then manipulate, have something appear to be very much liked by the public when in fact it is not, or promoted and retweeted when in fact it is not. This popularity is not real and essentially gain the system.

Thorold Barker: So if we take it back to where we started, if you look at the election that's coming up, how big a role will this big shift in AI capability over the last few months, which will obviously continue through the next year, how big an impact is this going to play, do you think in the messaging and the way that people get told the different pitches of the candidates?

Elon Musk: I think that's something we need to go and look at in a big way is to make sure that we're minimizing the impact of AI manipulation.

Thorold Barker: Okay, but beyond Twitter, are you worried about this for the election in general?

Elon Musk: Yeah, there probably will be attempts to use AI to manipulate the public and some of it will be successful and if not this election, for sure the next one.

Thorold Barker: We talk a lot in terms of AI about the next five to 10 years and what the impact is going to be on jobs and some of these things. If you look out on a much longer timeframe, given the speed and scale of the change and you look to your grandkids and great grandkids, can you just give us a sense of what it is going to be like to be human? How much is this going to change the fundamental nature of how we operate as a race at this point?

Elon Musk: I think it's going to change a lot, especially if you go further out into the future. I mean there will be, everything will be automatic. I mean there'll be household robots that you can fully talk to as though there are people that can help you around the house. There'll be a companion or whatever the case may be. There will be humanoid robots throughout factories and cars will also be all automatic and anything where intelligence can be applied, even moderate intelligence will be automated. So if you say like 10, 20 years from now.

Thorold Barker: Okay. But the actual broad thrust of, I mean jobs will change, but it'll be more AI enabling and making it better and easier rather than wholesale complete change of the skills you need.

Elon Musk: I mean, it depends about what timeframe we're talking about here. So if you say over 2030 year timeframe, I think things will be transformed beyond belief. You probably won't recognize society in 30 years. I do think we're fairly close. You asked me about artificial general intelligence. I think we're perhaps only three years, maybe six years away from it, this decade. So in fact, arguably, we are on the event horizon of the black hole that is artificial super intelligence.

Ryan Knutson: That's all for today, Wednesday, May 24th. The Journal is a co-production of Gimlet and The Wall Street Journal. If you like our show, follow us on Spotify or wherever you get your podcasts. We're out every weekday afternoon. Thanks for listening. See you tomorrow.

See the article here:

Elon Musk on 2024 Politics, Succession Plans and Whether AI Will ... - The Wall Street Journal

Read More..