The basic form of AI is “LLM,” or large-scale language model. This is an AI mechanism that learns “language patterns” by reading large amounts of text. BERT and GPT-2, which appeared around 2018, were the first examples of LLM, and were primarily used for research.
This evolved into the “GPT” series, a representative LLM model. GPT-3, released in 2020, brought major advances in text generation, and when it was released as ChatGPT at the end of 2022, it quickly became accessible to general users.
Now, the latest models, such as GPT-4 and Gemini, have appeared, and can even handle images and audio.

First, research support. This has two aspects. One is summarizing and organizing existing materials. The other is having AI perform the research itself, such as searching for reference cases, papers, and articles. This can significantly reduce the issue of “spending too much time on research.”

It is also beginning to be used in creative fields.
First, copy generation. Using AI, we can now quickly generate multiple ideas for advertising copy and social media posts.
Next, design. By using generative AI to generate image boards, rough drafts, and actual design comps, we can quickly discuss direction and deliver the actual designs.

This has resulted in significant benefits for our strategic planning service.

These include reduced work time, reduced document creation burden, the ability to produce multiple ideas at once, and making it easier for even new employees to produce a consistent output.

However, there are also risks that cannot be ignored.

Copyright issues: the possibility of someone else’s material being included in the generated results.

Information leakage: the risk that confidential information entered will remain in the cloud.

Accuracy variations: inconsistent answers to the same question, or factual errors.

Ethical concerns: the possibility of AI directly reproducing the biases it has learned.

In other words, understanding both the benefits and risks, and considering how much to delegate and what roles to assign to humans, is the key to using AI.

Tōsha ga sutoratejikkupuran’ningu o okonau sai wa itsumo kadai ga shōjite imashita. Shōninzū de habahiroi gyōmu o tantō shite iru tame, narejji no kyōyū ya chikuseki ga muzukashī. Shiryō sakusei ya chōsa ni jikan ga kakari, honrai no kikaku ya kurieitibu ni sakeru jikan ga kagira rete shimatte iru. Sarani kyōiku ya jinzai ikusei ni jūbun’na risōsu o saku no mo muzukashī. Kono jōkyō o AI de sapōto dekinai ka, to iu no ga shuppatsu-tendesu. Tatoeba sutoratejikkupuran’ningu no sakusei o okonau baai. Kako no sankō shiryō o sagasu no ni jikan ga kakari, fōmatto ya kōsei o ichi kara kangaenakereba narimasen. Jōhō seiri ni owa rete kanjin no aidea-dashi ga atomawashi ni naru koto mo ōi to omoimasu. Koko o AI ni makaseru koto de, sankō kikaku-sho o teiji shi tari, fōmatto o seisei shi tari to, tanjikan de shitaji o tsukureru yō ni shitai to kangaeta nodesu. Mata, tōsha ga kaihatsu shita `sutoratejikkufurēmuwāku: 6 E’ no nōhau mo sutoratejikkupuran’ningu ni wa hitsuyōdesu. `6 E’ wa, sutoratejikkupuran’ningu suruuede wa kihon-tekina jōhō to nari, sono pēji-sū wa sen pēji ni mo oyobimasu. Sokode AI o katsuyō shi,`6 e’ no jōhō to kako no sakusei shiryō o torikomi, aratana yōken (puronputo) o kuwaeru koto de, saishin no sutoratejikkupuran’ningu ga dekiru yō yō ni naru nodesu. Soryūshon zentai-zō wa kono yōna imējidesu. Kako no kikaku-sho ya manyuaru o narejjidētabēsu ni chikuseki shi, AI ga soko kara jōhō o kensaku yōyaku sai kōsei shimasu. Seika-mono wa pawāpointo ya Word to itta, genba de sugu tsukaeru keishiki de shutsuryoku sa remasu. Tsumari “dēta no chikuseki ￫ AI katsuyō ￫ jissai ni tsukaeru shiryō-ka” to iu nagare o ikkitsūkan de shien suru shikumidesu. < Img src =" http: / / Strategicpartners. Jp/ websys/ wp - kontentsu/ uploads/ 2025/ 09/ sukurīnshotto - 2025 - 09 - 08 - 18. 32. 35 - 750 X 331. Png" alt ="" width =" 750" haito =" 331" class =" alignnone size - medium wp - imāju - 81021" / > pointo wa, kurōzudo no dētadesu. Yoku senryaku no adobaisu o “GPT” ni kiku hito ga imasuga, koreha netto no sekai kara kikidashi shūyaku shita mono ni sugimasen. Shikashi, tōsha dokuji no nōhau o bēsu ni sureba, ōpundēta to kumiawaseru koto de, ebidensu no aru “kiku” autoputto ga kanō ni naru nodesu.
さらに表示
1,053 / 5,000
Our company has always faced challenges when undertaking strategic planning.

With a small number of staff handling a wide range of tasks, it’s difficult to share and accumulate knowledge.

Document creation and research take time, limiting the time available for actual planning and creative work.

It’s also difficult to allocate sufficient resources to education and human resource development.

Our starting point was to see if AI could support these situations.

For example, when creating a strategic plan, it takes time to search for past reference materials, and you have to think about the format and structure from scratch. I think it’s often the case that people are so busy organizing information that they put off coming up with ideas.

We wanted to entrust this process to AI, which could present reference proposals and generate formats, allowing us to lay the groundwork in a short amount of time.

Strategic planning also requires the know-how of our “Strategic Framework: 6e.”
“6e” is fundamental information for strategic planning and is over 1,000 pages long.

By utilizing AI, we can incorporate information from the 6e and previously created documents, and add new requirements (prompts), enabling up-to-date strategic planning.

The overall solution can be visualized as follows:
Past proposals and manuals are stored in a knowledge database, and AI searches, summarizes, and reconstructs the information.
The results are output in formats that can be used immediately on-site, such as PowerPoint or Word.
In other words, it is a system that supports the entire process of “data accumulation → AI utilization → production of usable documents.”

The key is closed data.
People often ask “GPT” for strategic advice, but this is simply a collection of information gleaned from the internet.
However, by combining our unique know-how with open data,
we are able to provide evidence-based, effective output.

AI has its “weak areas” and “bad habits.”
One is a phenomenon called “hallucination,” which causes it to tell plausible lies.
Second, output that lacks accuracy.
Even if a sentence is well-formed, the content can be inaccurate.
And unlike humans, it does not have the true “ability to think.” It simply predicts and outputs word patterns.
The important question here is, “Can AI or systems compensate for these weaknesses?” Or, “Do humans always need to take on this role?”

We are still a long way from the “AI that can do anything, like a human.”
Current generative AI is merely at the stage of “producing plausible words or images in response to specific input.”
Beyond that, concepts such as “artificial general intelligence (AGI)” and “artificial superintelligence (ASI)” exist, but we have not even reached AGI.

In other words, we need to consider how to interact with AI, keeping in mind that current AI is not omnipotent.

The ability to understand the emotions of customers and consumers, and the final decision on concept-making and expression.
Also, the perspective to determine whether something is ethically and socially appropriate.
And communication that builds trust with the team and customers.
These are values ​​that cannot be replaced by AI, and are uniquely human.

Even as we develop strategic planning AI, how do we deal with these “weak areas” and “bad habits”?
What should our staff training curriculum be like?
We believe that by meticulously pursuing these issues, our services will evolve even further.

Written by Sayaka Kaizu