This week, Claude 2 — the latest rival to OpenAI’s ChatGPT — was released by AI startup Anthropic.
Powered by a large language model, the chatbot will respond to questions and prompts in natural language. It can also summarize text, write topics, and assist in coding.
So, if ChatGPT is down and you’re scratching your head for another option, this “safe and steerable” alternative may fit the bill. Let's take a look at it.
Who's Behind This Ethical Chatbot?
Anthropic considers itself an AI safety and research company. Founded by former senior members of the OpenAI team, Daniela and Dario Amodei, it aims to be ethically driven in comparison to its generative AI counterparts and has a primary focus on “reliable, interpretable, and steerable” AI systems.
Claude 2 is the second version of its chatbot model, which had previously only been available to businesses via integration with products like Slack and Zoom. Anthropic claim that Claude 2 is easier to converse with than its predecessor and can produce even longer responses.
The risks of chatbots’ misinformation, hallucinations, copyright infringements, and completely made-up sources are raising more and more concerns. However, Anthropic has also said Claude 2 is “less likely to produce harmful outputs.” In a blog post it stated:
“We've been iterating to improve the underlying safety of Claude 2, so that it is more harmless and harder to prompt to produce offensive or dangerous output.”
Interestingly, it was reported that Google had invested in Anthropic back in February 2023.
How To Start Using Claude 2
For those in the US and UK, Claude 2 is available now in open beta. Simply head to Claude.ai and enter an email address or sign in with Google.
You’ll get a reminder that the chatbot is currently in beta and some of its features may not be available for free users. There will then be a handful of pages reminding users that the chatbot shouldn’t be used to give “legal, financial and medical advice,” absolving Anthropic and Claude 2 of any misuse.
It’s also worth remembering that some of your conversations with Claude 2 could be reviewed and used to improve Anthropic’s safety systems.
After all the formalities you’ll get to the chatbot itself, which features a clean and simple user interface. Once at this stage, you’re ready to get going with Claude 2, and Anthropic has even provided suggested prompts, like “summarize this PDF document in a bullet point outline.”
Plans to roll out the chatbot globally are in the works and should happen in the next few months.
Is Claude 2 Any Good?
New chatbots and AI text generators are all well and good, but if they don’t hit the mark performance-wise then their chances of success against the likes of ChatGPT and Bard are pretty slim.
According to Anthropic, Claude 2 scored 76.5% on the multiple-choice section of the Bar exam. It also features coding skills that outweigh its predecessor, scoring 71.2% on a Python coding test, compared to it's predecessor's 56% on the same test.
While it seems a legitimate competitor to the current AI heavyweights, having the most robust and intelligent performance is, of course, not at the top of the agenda for this “responsible and ethical” alternative. Anthropic has even created measures to ensure its chatbot is less vulnerable to corrupt uses.
“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, in a statement
While it’s a case of “time will tell” as to whether Claude 2 joins ChatGPT and Bard as the primary chatbots of choice, its introduction has just certainly things a lot more interesting.