While it appears that most of the think tank community is safe for now, artificial intelligence (AI) is getting so good that it may one day be able to replace think tanks altogether.
As Axios recently pointed out, the startup Primer is offering natural language processing (NLP) models for businesses that can rapidly read and analyze written text of all kinds. It also notes that businesses are creating digital workers out of software bots. Here are a few other things AI can now do:
- AI has the ability to display human-like qualities such as reasoning, learning, planning, and creativity. It can also see, hear, speak, smell, touch, move, and understand.
- AI can read text, write it, and even convert it into computer code (check out AI21 Labs' Jurassic-1 Jumbo and OpenAI's Codex). Here is some of the best AI writing-assistant software.
- The New York Times, Washington Post, Associated Press, and Reuters have used AI for years to generate content via natural language generation (NLG).
- AI can successfully debate humans in complex subjects (see IBM's Project Debater).
- Neural networks are providing automated feedback.
- Software like Synthesia allows one to create AI-generated videos from text in 40+ languages; it also allows you to create your own think tanker avatar.
Here is a piece on 21 uses of AI you probably didn't know about.
But all this new technology doesn't necessarily mean that think tanking will be dead soon. Here is an excerpt from a recent piece by National University of Singapore professor Atreyi Kankanhalli:
In my view, almost all...research tasks are currently not replaceable by AI. While AI can support search of references for literature review and discovery of patterns from data, it fails considerably in research problem identification and theory building, since these activities require semantic understanding that AI is currently not capable of. While AI could assist in data analysis, understanding the contributions of the work requires human interpretation. Similarly, for the writing process, AI mainly provides tools for preparing an initial publication draft – helpful in some science fields, which are more structured than humanities fields. Further, in the near future AI tools may not be able to replace our research activities because they lack semantic understanding, where little progress has been made so far.
The immediate impact of AI, for those who embrace it, will likely mean researchers can enhance their analysis with powerful tools to sift through enormous amounts of data and speed up a number of tasks.
In the coming years, Think Tank Watch will be establishing what is believed to be the first AI-run think tank to see if it can legitimately replace certain traditional think tanks. One major benefit is that the new think tank will not be biased by outside funders, a problem with nearly every major think tank to date.
As reported last year, there is even a new ranking of think tanks that uses AI to make its determination. There are also a number of think tanks that study AI. Georgetown University recently created the Center for Security and Emerging Technology (CSET), which has a big AI focus. CSET received a $55 million grant from the Open Philanthropy Project, a nonprofit research and grant-making group.
Here is a piece by Canon Institute for Global Studies (CIGS) Research Director Kunihiko Miyake entitled "In the age of AI, think tanks must evolve."
Here is a piece from Bruegel Director Guntram Wolff entitled "Artificial Intelligence: An Opporunity and a Challenge for Think Tanks."
And for any think tankers thinking about a new job, you can always check the site Will Robots Take My Job? to find out how susceptible your new job would be to computerization.
