OpenAI adds new features to ChatGPT including voice chat
After its latest update, ChatGPT can now access the internet to fetch real-time data and has the capability to recognize image inputs and process user speech via Whisper, an open-source speech recognition model.
According to OpenAI, “Browsing is particularly useful for tasks that require up-to-date information, such as helping you with technical research, trying to choose a bike, or planning a vacation.”
But, while internet access will allow a wider pool of resources to answer queries, it does not necessarily translate to improved accuracy. A recent study showed that more than half of the programming questions provided by ChatGPT were incorrect. Surprisingly, the users had an excessive amount of confidence in the answers.
AI search engines are prone to mistakes and users are advised to double-check the responses using the provided links. ChatGPT subreddit users have already expressed disappointment with the internet search feature.
OpenAI notes that the introduction of the new image prompt feature was highly influenced by its collaboration with Be My Eyes, including feedback on how to make the service most useful for sight-impaired users.
With these new capabilities, users can easily convert flow charts into code, troubleshoot errors in circuit diagrams, or even upload a picture of a PDF for the AI to create a quick summary.
The new voice model, on the other hand, can also translate content and client input for businesses. This potentially improves communication with the customers and enhances engagement. Another possibility would be for the firms to use the voice capabilities to translate the voice of the speaker in various supported languages.
Spotify has already shown that it’s possible to retain the voiceprint of the original host when translating for podcasts. Spotify’s Voice Translation feature is the result of a collaboration with OpenAI.
Another major advantage of the new GPT-4 voice processing is that chatGPT is now accessible even to people who can’t use a keyboard. Hopefully, this will help lower the cost of accessibility tech which usually has a price tag that’s unreachable for most people.
The newly-updated ChatGPT can now comfortably compete with rivals, Google Bard and Bing Chat. It’ll especially be interesting to see how ChatGPT compares with Bing Chat seeing as they both use GPT-4 generative AI search.