The public can now purchase Claude 2 AI, a ChatGPT rival from Anthropic. The company claims that it is better than its predecessor in many ways that users will benefit from. But the real question that we need to be asking is, “Is it good enough to take ChatGPT’s throne?”
Daniela and Dario Amodei, two former senior members of the OpenAI team, established Anthropic. According to its website, the startup aims to make generative AI safe and “steerable” by focusing on ethics. Notably, news of Google’s investment in the AI business surfaced in February. The second iteration of Claude, which was previously solely accessible to enterprises, is the large language model.
The successor to Anthropic’s initial commercial model, Claude 2 AI, is now accessible in beta in the United States and the United Kingdom via a paid API (with limited access) and the web. The cost of the API hasn’t changed (around $0.0465 to produce 1,000 words), and Claude 2 is already being tested by a number of companies, including Sourcegraph and the generative AI platform Jasper.
Meet xAI, Elon Musk’s new AI company against OpenAI
Claude was initially released in March by Google-backed Anthropic, but organizations could only access it by request or via a Slack app. Anthropic is enhancing the chatbot’s capabilities with Claude 2 by making a number of changes. Claude 2 is slightly better in math, coding, and reasoning than the original Claude model, in addition to being able to construct longer replies.
What is Claude 2 AI?
Claude 2 AI is the second-generation artificial intelligence tool developed by Anthropic, a company founded by ex-OpenAI employees. Claude is aimed to be “safer” than other AI tools out there.
According to reports, Claude 2 is a powerful AI that can process up to 100,000 tokens—roughly equal to 75,000 words—in a single request. This represents a significant increase over Claude’s prior 9,000 token cap, which offers a special benefit: the AI will now be able to respond in a more improved and contextual manner.
Anthropic asserts that Claude 2’s behavior is easier to comprehend and easier to modify as necessary when compared to other models because of constitutional AI. However, the business also accepts that there are other training methods besides constitutional AI.
According to Anthropic, many of the guiding principles for Claude 2 were established through a “trial-and-error” method, and it has been necessary to make numerous tweaks to stop its models from being overly “judgmental” or “annoying.”
However, unlike Bard and Bing, Claude 2 is still not online and is trained using data as of December 2022. While this prevents it from providing the most latest information on current affairs, its dataset is still more recent than the one used by ChatGPT’s free edition. ChatGPT was trained until 2021.
Claude’s context window has also just been increased by Anthropic to about 75,000 words. The bot can parse dozens of pages or even an entire novel if you upload them to it.
Therefore, Claude is your bot if you require a brief overview of challenging and lengthy study work. In contrast to other models, ChatGPT has a maximum word count of about 3,000. I’m eager to try Anthropic now that it’s open to the general public and see if a larger context window can, as we saw with Bing, derail this “harmless” bot.
Luckily, Claude 2 AI is completely free, and you don’t have to pay a single dime to benefit from its features, technologies, and innovations. It is probably one of the best reasons for thousands of users.
However, you can also try the Claude 2 AI API, which costs ~$0.0465 to generate 1,000 words.
Currently, Claude 2 is only available for users in the United States and the United Kingdom. If you are residing in a different country, you will have to wait longer to try out the latest AI content generator.
Some people use VPN services to try these kinds of new technologies that launch exclusively for Western countries. That is an option that many people use in similar scenarios but if you wait a little longer, you won’T have to deal with any VPN hassle as it will probably come to your country too.
How to use Claude 2 AI
Visit the Claude 2 page on Anthropic’s website to sign up for free if you’re in the US or the UK. When you simply click “Talk to Claude,” a request for your email address will appear. You’ll be prepared to depart once you’ve verified the address you entered.
Claude 1.3 vs. Claude 2 AI
Like the original Claude (Claude 1.3), Claude 2 is capable of document-wide searches, summaries, writing and coding, and question-and-answer sessions on specific subjects. Anthropic asserts that Claude 2 is superior in a number of ways.
For instance, Claude 2 performs marginally better than Claude 1.3 on a multiple-choice portion of the bar exam (76.5 percent versus 73.3 percent). It is capable of passing the U.S. Medical Licensing Exam’s multiple-choice section. Additionally, it is a better programmer than Claude 1.3, scoring 71.2% on the Codex Human Level Python coding test.
Additionally, Claude 2 is 2.8 percentage points better in math than Claude 1.3, answering 88% of grade-level questions correctly on the GSM8K collection. Claude 2 AI is also two times better than Claude 1 at “giving harmless responses,” according to Anthropic. In comparison to the prior model, it should therefore be less likely to spit out hazardous stuff when you engage with it, albeit Anthropic does not completely rule out the potential of jailbreaking.
ChatGPT vs. Claude 2 AI
Claude handles large amounts of data but may not be the best choice for tasks requiring real-time data. ChatGPT is more creative, which is perfect for tasks requiring specific language support.
The test results of Claude 2 are equivalent, albeit slightly lower, than those of GPT-4, which powers the flagship version of ChatGPT. But Anthropic aims to distinguish itself from OpenAI by offering a more moral and accountable substitute.
“We have an internal red-teaming evaluation that scores our models on a large representative set of harmful prompts, using an automated test while we also regularly check the results manually,” Anthropic said in the announcement.
It is proven that Claude 2 AI is better than its own predecessor, but ChatGPT will probably continue to be preferred by a vast amount of users as even though Claude 2 has some advantages, it still doesn’t have the whole package to take down the king. It looks like it will be hard for other companies to take down the GPT kingdom, as even former OpenAI employees can’t get near it.
On another note, you don’t have to choose one between ChatGPT, Claude 2 AI, Google Bard, or Microsoft Bing. Every tool has different advantages and features that make them stand out against others. IF you want to benefit more from AI tools, use all of them according to your needs and preferences.