The past 3 months have been “the iPhone moment” for Artificial Intelligence with the release of chatGPT in December 2022. Yesterday, 14 March saw the release of GPT4, the next iteration of OpenAI’s large language model.
GPT4 has many new abilities, some of which will only be uncovered as it is used widely. The 3 highlighted in openAI’s release notes are:
- Ability to understand both images and text as input (though it is limited for now to text as output)
- Much greater performance in term of complex comprehension tasks as measured by college entrance exams and other standard tests
- Finally, it is much more proficient in other languages and showed surprisingly good performance in Afrikaans and Swahili in the release notes
All of this shows a marked improvement in the intelligence of the tool as it becomes closer to human-level ability.
However, it still suffers from many problems such as ‘hallucinating’ false answers.
There is certainly a lot of hype but we are seeing companies investing and adopting at unprecedented rates around the world. Our teams are already working on implementation projects. Managing the ethical and responsible use of it will be key to be successful.
We are also seeing wide-spread adoption through ’shadow AI’ – staff, knowledge workers in particular, using the tool unofficially in their day to day job – to draft emails, write parts of reports, prepare for presentations etc.
Importantly, we are seeing boards and executive teams start to wrestle with the strategic implications – as the marginal ‘cost of intelligence’ (previously something that only humans could provide) becomes cheaper and cheaper.
Companies who have not engaged with it yet should start to educate their executive teams and boards.
We are still early on the adoption journey and the tech itself will continue to evolve so they should stay informed and become more ‘curious than convicted’.