With the advent of generative AI and LLMs, the era of online search was destined for a revolution. AI transformed blue-colored lists of links into direct answers. We saw it with Perplexity, we saw it accidentally with Google itself, and now we see it with iAsk AI.
Why it matters. iAsk beats traditional search engines and AI giants like ChatGPT in some synthetic performance tests. It’s in its destiny to make a name for itself.
iAsk AI in action. Image | Xataka
In detail. iAsk AI debuted over a year ago and already processes around 1.5 million daily searches. This number doesn’t compare to Google’s 8.5 billion, but it’s the beginning of something more.
iAsk AI uses advanced natural language processing to understand the context and intent behind each search rather than just the keywords. According to benchmarks:
- It has an accuracy of 93.9% on MMLU, a multiple-choice benchmark covering 57 knowledge areas.
- iAsk AI outperformed GPT-4 by over 10 percentage points.
- This model achieved 90.1% accuracy on TruthfulQA, a benchmark designed to measure the truthfulness of linguistic models in generating answers in 38 domains, compared to 59% accuracy on GPT-4.
iAsk Pro, its advanced and paid option, leads the MMLU Pro benchmark, which experts consider the gold standard for evaluating AI language models. MMLU Pro covers 14 major knowledge areas, broken down into more than 12,000 questions from textbooks and academic exams.
Between the lines. This milestone positions iAsk AI as a search engine that users should consider. It also demonstrates that this AI model has the potential to be the most accurate among those available today.
However, it must prove itself in everyday, large-scale use, something Google has been perfecting for decades.
Another practical example. We asked iAsk, “How would you resolve a complex ethical conflict, such as the trolley dilemma, by explaining the possible consequences of each option from different philosophical perspectives?
The answer details the utilitarian, virtuous, deontological, and social contract perspectives. It also lists each perspective’s positive and negative consequences and refers to the sources it used.
The contrast. Google specializes in getting reliable results from keywords using extensive web coverage. iAsk offers direct answers and superior contextual understanding.
Currently, Google seems better at finding specific web pages, but iAsk is on another level when providing direct answers.
The bottom line. iAsk has a lot of ground to gain as it competes with the search giant Google and others, starting with Perplexity. It uses similar technologies and has similar goals.
Now, it remains to see how far it can go as a search engine or whether it can maintain its advantage in benchmarks against Google and OpenAI and translate this into a user experience that captures consumers.
Images | Pixabay edited by Xataka On
View 0 comments