Claude 2 is the latest ChatGPT rival you should consider incorporating into your daily life. Like ChatGPT, Claude 2 is available for free to users, but it might have one advantage over OpenAI’s software out of the gate. As I explained before, Claude 2 is the equivalent of ChatGPT’s GPT-4 model, which is only available to users who pay the $20 ChatGPT Plus monthly subscription.
Claude 2 is also supposed to be a more ethical generative AI, built on top of a constitution that should protect the user. That’s not to say that OpenAI, Microsoft, or Google do not care about the user. Or that Claude 2 won’t surface hallucinations in its responses. But Anthropic, the makers of Claude 2, might care more about the ethical aspect of AI.
On top of that, Claude 2 does have a few big advantages over ChatGPT, which should matter to anyone not looking to pay for a premium generic AI experience.
Access newer information from the web
Recently, I used ChatGPT to buy new running shoes that fit several requirements, and I did it fully, knowing that OpenAI’s data set isn’t newer than September 2021. ChatGPT isn’t connected to the web, so it offered me suggestions relevant as of two years ago.
Still, I used that data to find newer versions of those shoes and make my choice. ChatGPT was definitely helpful, though I would have preferred it to offer me more recent information.
Sadly, Claude 2 isn’t connected to the live internet either. But it’s trained on a newer data set, going up to December 2022. That’s over a year of extra data for Claude 2, which can be a massive advantage. You won’t have to rely on traditional search for information that ChatGPT can’t offer if you switch to Claude 2.
As more time passes, both services will need upgrades.
Upload files to summarize automatically
Another massive Claude 2 advantage over ChatGPT is the support to upload files to the Anthropic servers. Claude 2 has an attachment button that lets you upload files of up to 10MB. You can add up to five files with each prompt.
Claude 2 will then summarize the files without requiring any prompts. That’s another great advantage over ChatGPT and a great productivity hack. The chatbot will then be able to continue the conversation about the contents of those documents as you input additional prompts.
That’s a great way to explore documents that might be more difficult to understand or take longer to read. Like a study or a technical paper.
Process more words than ChatGPT
Generative AI products might offer free access to users, but they cost money to run. And that’s why they have limitations on the number of words they can process at a time. ChatGPT supports prompts of 8,000 tokens, 12.5 times fewer than Claude 2.
With Claude 2, you can use 100,000 tokens in a prompt, amounting to about 75,000 words per prompt. You might never have to issue such a lengthy command to generative AI products. But if you want them to summarize larger pieces of content, Claude 2 might be a better option than ChatGPT, as it would do it faster.
Read links to online articles
Again, Claude 2 isn’t connected to the live internet. Neither is ChatGPT. But you can feed the former links to online content, and the chatbot will read them. That’s another huge advantage of Claude 2 over OpenAI’s service, one that we’ve highlighted before.
You don’t need the chatbot you’re engaging with to be connected to the web if it can read links you include in prompts.
Bypass paywalled content?
You should be aware of hallucinations when Claude accesses content through links, as ZDNet explains. And, again, all chatbots hallucinate right now. That’s why products like Google SGE are available in beta only.
What’s strange, considering Anthropic’s ethical approach to generative AI, is that Claude 2 was able to bypass paywalled content in ZDNet’s test:
The fact that it seems like it can access content through links brings up the question of whether or not Claude can bypass paywalls, a characteristic that could infringe copyright laws. I did give Claude paywalled links to test its capabilities. Claude gave a summary that was similar to the answer it gave using the copied text, albeit with hallucinations.
If this is a Claude 2 feature, it’s certainly one that Anthropic will fix. Infringing on copyright content is another big issue generative AI software developers have to prevent.
But it might be another thing Claude 2 can do that ChatGPT can’t. Well, you can’t test it on OpenAI’s software since you can’t feed the AI links to paid content.
